diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..ee623177606ec2d3382885785a58af6221a4b6ee Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..1f21d6a3c6167786d67fa0e2410af0957c84eb06 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..f746f461900fa47c926ca8e30a188bf824262b94 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-00616f328a8b4b5a8e689f61e70758b6-000/030/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-00616f328a8b4b5a8e689f61e70758b6-000/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..ae600b12455a20e55374f3b23b76e116b64e934b Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-00616f328a8b4b5a8e689f61e70758b6-000/030/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-00616f328a8b4b5a8e689f61e70758b6-000/060/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-00616f328a8b4b5a8e689f61e70758b6-000/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..869bf720b0e71d19f68351bae6dbad2ec5caec97 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-00616f328a8b4b5a8e689f61e70758b6-000/060/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-011005.txt b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-011005.txt new file mode 100644 index 0000000000000000000000000000000000000000..5faaade523a0f4b2759ed7f2d8e1710d233d7630 --- /dev/null +++ b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-011005.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-000-00616f328a8b4b5a8e689f61e70758b6-000 --func_name main1 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-011154.txt b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-011154.txt new file mode 100644 index 0000000000000000000000000000000000000000..df62ec2af28521fb5ceb4c1c7561b898c42ef8f9 --- /dev/null +++ b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-011154.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-000-00616f328a8b4b5a8e689f61e70758b6-000 --func_name main23 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..6f8fd0cf75cd09997e7111a9909d09677c101624 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..6e06501fef6e8f0fe118106688923eaff5f0acf4 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..0149979f92381a947b8b5cf67898641c4d3a2632 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..948fcbfb9296e09eb80de8441b0253c15539f71c Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..105d2e5e670a79ff675e8eeb2235a51b3eb0254a Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..18ca9dd861c05d1be0f603684f762732e0aeaa55 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..d0f9ab3851aa1de47fa697c219a82ebac5688c41 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..0f504da63a9e8f835c1b3c074cf9673edb1d8949 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..33654ea4e6eb3e246828b4c1d8a68d51098c7223 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..0ae86c1be6586b43dfbad2ae16c150afec1e278c Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..1096e2bd4486dce6f921fe070e95b7e46692b1e1 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt new file mode 100644 index 0000000000000000000000000000000000000000..58be10068dfa45d0788494b12923f56764e31e8e --- /dev/null +++ b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt @@ -0,0 +1,6 @@ +Function Execution Times +======================== + +photometric_reconstruct_from_pretrained: 1755.3891 seconds + +Total execution time: 1755.3891 seconds diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..d8bc23609e3a8e259e979852b45bf3d0115c2b85 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..6bab05692bf85ef04346065ecb5e88ab2f451bba Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..e4f7175c122103ea57f85e41cfdd99e4be3eb711 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..fb3ec282a22ad7c12a0bdcb630b34b16946fede6 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2392bf97ccdb2256b325306a1de8e64ca248162f Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2229b7f4f22e291970c693cc52f7c00f1b40ec16 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..43a8ab1273c5dee3b30db05539dfd383451cc779 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..5619e4538b9b60b4f461a2229ac798395d8c4b3e Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..4858363d4950cc29d6de9b88a338e01a4e224519 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..d20bdbc1dbf1416bf6d91b591a49948a373cfb95 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..b95a4d0fa9ece67e9ee66b83c5ae2cc504f06ac4 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_45/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_45/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..13b4f522dd430a0f76a8e6063a6e3db4d9a24f67 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_45/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..7bb6f9dd4bdede7430db976e61a7e2d9dbaf31f2 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..7346c513c3d7f8e17e6a9b53e45db05fb2f95294 Binary files /dev/null and b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx differ diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt new file mode 100644 index 0000000000000000000000000000000000000000..6a1fec3cfc60bbfaa70f28eacb245561c8a62af9 --- /dev/null +++ b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt @@ -0,0 +1 @@ +FPS: 170.3734385564 diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py new file mode 100644 index 0000000000000000000000000000000000000000..7016b5610ec6e7552d04c0af22d9b5b124beafb4 --- /dev/null +++ b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py @@ -0,0 +1,2928 @@ +import sys +import os +sys.path.append(os.path.dirname("../lib_render")) +sys.path.append(os.path.dirname("../lib_mosca/gs_utils")) +sys.path.append(os.path.dirname("..")) + +import torch +import numpy as np +from scipy.spatial.transform import Rotation +# import matplotlib.pyplot as plt +import pickle +import random +import glob + +# import pypose as pp +from torch import nn + +from itertools import chain +from scipy.spatial import KDTree,Delaunay,distance_matrix +import open3d as o3d +from collections import defaultdict +import roma + +from mpl_toolkits.mplot3d import Axes3D + +# import hdbscan +from sklearn.cluster import DBSCAN +from tqdm import tqdm + +from lib_mosca.mosca import __DQB_warp2__, __LBS_warp2__ + +from lib_mosca.gs_utils.utils_helper import project_points, draw_projected_points, draw_projected_points_masked, valid_and_visible, overlay_mask_yellow, stack_images_side_by_side, compute_depth_gradient, associate_pts_value, draw_projected_points_colored +from lib_mosca.gs_utils.som_vis_utils import ( + apply_depth_colormap, +) + +import matplotlib.pyplot as plt +import argparse +import logging, glob, sys, os, shutil, os.path as osp + +def get_order(arr): + """ + Computes the rank (order) of each element in the array. + + Args: + arr (np.ndarray): Input 1D array. + + Returns: + np.ndarray: Array where the i-th element is the rank of the i-th element in the input array. + """ + # Get the indices that would sort the array + sorted_indices = np.argsort(arr) + + # Create an array to store the ranks + order = np.empty_like(sorted_indices) + + # Assign ranks + order[sorted_indices] = np.arange(len(arr)) + + return order + +def count_out_dict_files(folder_path): + # Use glob to find files matching the pattern + file_pattern = os.path.join(folder_path, 'out_dict_*.pkl') + files = glob.glob(file_pattern) + return len(files) + +def calculate_edge_relative_orientation(Ts): # 854 pts will use ?? GB memory!!! + nk=Ts.shape[1] + Ts1=Ts.unsqueeze(2).repeat(1,1,nk,1) + Ts2=Ts.unsqueeze(1).repeat(1,nk,1,1) + Ts12=(Ts1.Inv()*Ts2) + # Ts_rel=relativelieAlgebra_pp(Ts1,Ts2) + return Ts12 + +def get_navie_ori_distances(Oris_rel): # TODO + Oris_rel_so3=Oris_rel.Log() # (nt, nk, nk, 3) + norm_oris_rel_so3=torch.norm(Oris_rel_so3.tensor(),dim=-1) # (nt, nk, nk) + norm_ori_rel_so3=torch.mean(norm_oris_rel_so3,axis=0) # (nk, nk) + return 1.0 - torch.cos(norm_ori_rel_so3) # (nk, nk) + + +def compute_accel_loss(transls): + # transls [nt,n,3] or oris [nt,n,4] + accel = 2 * transls[1:-1] - transls[:-2] - transls[2:] # [nt-2,n,3/4] + loss = accel.norm(dim=-1).mean() # [nt-2,n,3/4] => [nt-2,n]=> [1] + return loss +def cal_rigidity_loss(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global + stats = { + "loss_local_rigidity": loss_local_rigidity, + "loss_local_rigidity_ori": loss_local_rigidity_ori, + "loss_local_global": loss_local_global, + } + + return loss, stats + +def cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges): + oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + return loss_local_rigidity_dt,loss_local_rigidity_ori_dt + +def cal_rigidity_loss_extra(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + nt=transls.shape[0] + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + # 4. compute dt rigidity loss + # delta_t=50 + # oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + # oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + # transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + # transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + # dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + # error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # # 2. compute local rigidity loss for oris + # qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + # qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + # qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + # error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + if nt<100: + delta_t=int(nt/2) + else: + delta_t=50 + loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + # loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(50,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss_local_rigidity_10,loss_local_rigidity_ori_10=cal_dt_loss(10,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global+100*loss_local_rigidity_50+100*loss_local_rigidity_10#+100*loss_local_rigidity_ori_dt + stats = { + "loss_local_rigidity": loss_local_rigidity.item(), + "loss_local_rigidity_ori": loss_local_rigidity_ori.item(), + "loss_local_global": loss_local_global.item(), + "loss_local_rigidity_50": loss_local_rigidity_50.item(), + "loss_local_rigidity_10": loss_local_rigidity_10.item(), + } + + return loss, stats + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +# optimize_overall_with_orientation version of +def optimize_overall_with_orientation_edges( + edges, + Transls, Transls_optimizable, Mask, Dists_optimizable, weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, Ori_distances_optimizable, weight_ori_distance_optimizable, + Dists_observed_mean, Dists_observed_std, Mask_observed,Nearby_mask, + W2Cs,Depths_info_perG_key, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 ###### test + + # Depths_info_perG_key [nt,nk] + + nt = Transls.shape[0] + nk = Transls.shape[1] + ne = edges.shape[0] + + # Device setup + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + Dists_optimizable = nn.Parameter(Dists_optimizable.to(device)) + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Ori_distances_optimizable = nn.Parameter(Ori_distances_optimizable.to(device)) + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Mask = Mask.to(device) + Depths_info_perG_key=Depths_info_perG_key.to(device) + + optimizer = torch.optim.Adam( + [Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable, Ori_distances_optimizable], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, threshold=1e-3, min_lr=min_lr + ) + + # Distance matrix helper function + def distance_matrix(x, y): + return torch.cdist(x, y, p=2) + + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt,3,3] + # I0=torch.eye(3,device=device) + # I0[2,2]=0.1 # 0.01 camel + # I0=I0[None,:,:] + # I1=W2Cs_R.transpose(-1, -2)@I0@W2Cs_R # [nt,3,3] + # I1=I1[:,None,:,:] + I0=torch.eye(3,device=device).expand(nt,nk,-1,-1).clone() # [nt,nk,3,3] + I0[...,-1,-1]=Depths_info_perG_key # 0.01 camel # CHECK TODO + I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nk,3,3] + # breakpoint() + + # Optimization loop + # for epoch in range(max_epochs): + for epoch in tqdm(range(max_epochs), desc="Training Epochs", unit="epoch"): + optimizer.zero_grad() + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + a1=(Transls_optimizable-Transls)[:,:,None,:] # (nt, nk, 1, 3) + a1_T=a1.transpose(-1, -2) + squared_norm=(a1@I1@a1_T).squeeze(-1,-2) + # squared_norm=(a1@a1_T).squeeze(-1,-2) # test + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # add epsilon for numerical stability + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) # full model + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) # ablation of loss + # Dist_confident_pts_error_masked=Dist_confident_pts_error[Mask] + # Dist_confident_pts_error_masked_disturbed=Dist_confident_pts_error_masked+torch.randn_like(Dist_confident_pts_error_masked)*0.01 + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed) + # mask_shuffle=generate_ratio_binary_tensor(n=Dist_confident_pts_error_masked_disturbed.shape[0], ratio_0=0.8) + # mask_shuffle=torch.ones_like(Dist_confident_pts_error_masked_disturbed,dtype=torch.bool) + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed[mask_shuffle]) + + + # Ori_confident_pts_error=torch.norm(Oris_xyzw_optimizable_normed-Oris_xyzw,dim=-1) # test + # loss_oris_confident_pts = torch.mean(Ori_confident_pts_error[Mask]) + + mask_edge=generate_ratio_binary_tensor(n=edges.shape[0], ratio_0=0.3) + edges_mask=edges[mask_edge] + weight_normed=torch.abs(weight_distance_optimizable)/torch.norm(weight_distance_optimizable,dim=-1,p=1)[:,None] # (nu,nk) + weight_edges=weight_normed[edges_mask[:,0],edges_mask[:,1]] # (ne) + dis_local_canonical_edge=Dists_optimizable[edges_mask[:,0],edges_mask[:,1]] # (ne) CHECK order + # loss_rigidity,stats_rigidity=cal_rigidity_loss(Transls_optimizable,Oris_xyzw_optimizable_normed,edges,weight_edges,dis_local_canonical_edge) + loss_rigidity,stats_rigidity=cal_rigidity_loss_extra(Transls_optimizable,Oris_xyzw_optimizable_normed,edges_mask,weight_edges,dis_local_canonical_edge) + + # Loss 3: Regularization on weight distance + loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch[:-1] - Dists_epoch[1:])) # (nt-1,nk,nk) + # Dists_epoch_edges2=Dists_epoch_edges.reshape(nt,ne) # (nt,ne,nk) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch_edges2[:-1] - Dists_epoch_edges2[1:])) # (nt-1,nk,nk) + # loss_local_smoothness = torch.mean(torch.abs(Transls_optimizable[:-1]-Transls_optimizable[1:])) # (nt-1,nk,nk) # [TODO] + loss_accel = compute_accel_loss(Transls_optimizable) + + + # Loss 7: symmetry loss for edge weights + # TODO + # loss_symmetry=torch.mean(torch.abs(weight_distance_optimizable-weight_distance_optimizable.transpose(0,1))) + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + # a_loss_oris_confident_pts=1*loss_oris_confident_pts + a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point 0.1 common value + a_loss_sparsity=1e-1/loss_sparsity + a_loss_accel=0.1*loss_accel + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # flag_train_orientation=True + # if flag_train_orientation: + # a_loss_density=5e-3/loss_sparsity # 1e-3 for pos+rotation + # else: + # a_loss_density=5e-4/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + # a_loss_oris_distance_confident_pts=1*loss_oris_distance_confident_pts + # a_loss_ori_dist=1*loss_ori_dist + # a_loss_ori_local_smoothness=0.01*loss_ori_local_smoothness + + total_loss = a_loss_dist_confident_pts + a_loss_rigidity + a_loss_accel # + a_loss_sparsity + # + a_loss_dist + a_loss_local_smoothness + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness + a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # + a_loss_oris_confident_pts + + # Backpropagation and optimization step + total_loss.backward() + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # Logging progress + rigidity1=stats_rigidity["loss_local_rigidity"] + rigidity_ori2=stats_rigidity["loss_local_rigidity_ori"] + global3=stats_rigidity["loss_local_global"] + if epoch % log_interval == 0 or epoch == max_epochs - 1 or optimizer.param_groups[0]['lr'] < min_lr * 2: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + # f"a_loss_density: {a_loss_density.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + # f"OConfDist: {a_loss_oris_confident_pts.item():.6f}, " + f"a_ridigity: {a_loss_rigidity.item():.6f}, " + # f"rigidity1:{rigidity1:.6f}, " + # f"rigidity_ori2:{rigidity_ori2:.6f}, " + # f"global3:{global3:.6f}, " + + # f"Dist: {a_loss_dist.item():.6f}, " + # f"OConfDist {a_loss_oris_distance_confident_pts.item():.6f}, " + # f"ODist {a_loss_ori_dist.item():.6f}, " + # f"Smooth: {a_loss_local_smoothness.item():.6f}, " + # f"OSmooth: {a_loss_ori_local_smoothness.item():.6f}, " + # f"Rigidity L: {loss_local_rigidity_2.item():.6f}, " + f"a_loss_accel: {a_loss_accel.item():.6f}, " + f"a_loss_sparsity: {a_loss_sparsity.item():.6f}, " + # f"Symmetry L: {a_loss_symmetry.item():.6f}, " + # f"W L: {loss_weight_distance.item():.6f}, " + # f"W L2: {loss_weight_distance_2.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + print(f"weight_normed: {(weight_normed>0.1).sum().item():.6f}") + # print(stats_rigidity) + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and epoch > 5000: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + return Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable_normed, Ori_distances_optimizable + + +def get_knn_edges(n_neighbors,Transls,Contribs,Transls2,Contribs2,contrib_threshold, flag_remove_self=False): + # choice 2: knn graph + def o3d_knn_p(pts2,pts, num_knn, flag_remove_self=False): + indices = [] + sq_dists = [] + pcd = o3d.geometry.PointCloud() + pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + pcd_tree = o3d.geometry.KDTreeFlann(pcd) + for p in pts2: + if not flag_remove_self: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + indices.append(i[0:]) + sq_dists.append(d[0:]) + else: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn + 1) + indices.append(i[1:]) + sq_dists.append(d[1:]) + return np.array(sq_dists), np.array(indices) + + nk=Transls.shape[1] + nu=Transls2.shape[1] + + i_s_knn_others,i_s_knn_is=[],[] + # n_neighbors=5 + for i in range(nu): + t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + pts2=[pt2] # just one point + sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors, flag_remove_self) + sq_dists,indices= sq_dists[0],indices[0] + i_s_knn=i_s[indices] # usage: ids[i_s_knn] + i_s_knn_others.append(i_s_knn) + i_s_knn_is.append(i*torch.ones_like(i_s_knn)) + i_s_knn_is=torch.cat(i_s_knn_is) + i_s_knn_others=torch.cat(i_s_knn_others) + edges_others=torch.stack([i_s_knn_is,i_s_knn_others],dim=-1) + # breakpoint() + # old version: has bug when knn key points are not enough + # i_s_knn_others=torch.stack(i_s_knn_others) + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # edges_others.shape # nu*5 + return edges_others + +# def get_knn_edges_naive(n_neighbors,Transls,Transls2, flag_remove_self=False): +# nk=Transls.shape[1] +# nu=Transls2.shape[1] +# nt=Transls.shape[0] +# Dist=[] +# for t in range(nt): +# dist=(Transls2[t][:,None]-Transls[t][None]).norm(dim=-1) +# Dist.append(dist) +# Dist=torch.stack(Dist,dim=0) +# Dist=Dist.mean(dim=0) + +# def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): +# """ +# Naive temporal k-NN graph construction using average distance over time. + +# Args: +# n_neighbors (int): Number of neighbors. +# Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. +# Transls2 (torch.Tensor): (T, N2, D) tensor of query points. +# flag_remove_self (bool): If True and N1 == N2, removes self-matches. + +# Returns: +# torch.Tensor: (N2 * n_neighbors, 2) edge index pairs [Transls2_index, Transls_index]. +# """ +# T, N1, D = Transls.shape +# N2 = Transls2.shape[1] + +# # Compute temporal average pairwise distances +# dist_list = [] +# for t in range(T): +# dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) +# dist_list.append(dist) +# Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + +# if flag_remove_self and N1 == N2: +# diag = torch.arange(N2, device=Dist.device) +# Dist[diag, diag] = float('inf') + +# # Get top-k nearest neighbors for each Transls2 point +# _, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + +# # Build edge index pairs +# row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) +# edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + +# return edges_others + +def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): + """ + Naive temporal k-NN graph construction using average distance over time. + + Args: + n_neighbors (int): Number of neighbors. + Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. + Transls2 (torch.Tensor): (T, N2, D) tensor of query points. + flag_remove_self (bool): If True and N1 == N2, removes self-matches. + + Returns: + edges_others (torch.Tensor): (N2 * k, 2) edge index pairs [Transls2_index, Transls_index]. + dists (torch.Tensor): (N2 * k,) mean distances corresponding to each edge. + """ + T, N1, D = Transls.shape + N2 = Transls2.shape[1] + + # Compute temporal average pairwise distances + dist_list = [] + for t in range(T): + dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) + dist_list.append(dist) + Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + + if flag_remove_self and N1 == N2: + diag = torch.arange(N2, device=Dist.device) + Dist[diag, diag] = float('inf') + + # Get top-k nearest neighbors and distances + dists, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + + # Build edge index pairs + row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) + edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + dists_edges = dists.reshape(-1) # (N2 * k,) + + return edges_others, dists_edges + + + + + +def optimize_others_with_orientation_edges4( + edges, + Transls, Transls_optimizable, Mask, #Dists_optimizable, #weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, #Ori_distances_optimizable, #weight_ori_distance_optimizable, + Transls_key, Oris_xyzw_key,Transls_key_optimized, Oris_xyzw_key_optimized, + weight_param_key_optimizable,sq_dists_key_nonkey_edges, + ts_most_confident,per_nonkeyGaussian_dweight, + W2Cs,Depths_info_perG_nonkey, + device, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 + # weight_param_key_optimizable [nk] per-keyGuassian weight + # Depths_info_perG_nonkey [nt,nu] + # nt = ts.shape[0] + + + nt_all = Transls.shape[0] + nu=Transls.shape[1] + nk=Transls_key_optimized.shape[1] + ne=edges.shape[0] + + Oris_xyzw=Oris_xyzw/Oris_xyzw.norm(dim=-1,keepdim=True) # normalize + # Oris_xyzw_optimizable=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) # normalize + Oris_xyzw_key_optimized=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) # normalize + + # Device setup + # Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + # Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Transls_optimizable=Transls_optimizable.to(device) + Oris_xyzw_optimizable=Oris_xyzw_optimizable.tensor().to(device) + + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Transls_key = Transls_key.to(device) + Oris_xyzw_key = Oris_xyzw_key.to(device) + Transls_key_optimized = Transls_key_optimized.to(device).detach() + Oris_xyzw_key_optimized = Oris_xyzw_key_optimized.to(device).detach() + weight_param_key_optimizable = nn.Parameter(weight_param_key_optimizable.to(device)) + # sq_dists_key_nonkey = sq_dists_key_nonkey.to(device) + sq_dists_key_nonkey_edges = sq_dists_key_nonkey_edges.to(device) + per_nonkeyGaussian_dweight = nn.Parameter(per_nonkeyGaussian_dweight.to(device)) # don't use for now, just keep + # per_nonkeyGaussian_dweight=None + Depths_info_perG_nonkey = Depths_info_perG_nonkey.to(device) + + # sq_dists_key_nonkey_edges=sq_dists_key_nonkey[edges[:,0],edges[:,1]] # (ne) + exp_sq_dists_key_nonkey_edges=torch.exp(-sq_dists_key_nonkey_edges) # (ne) + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt_all,3,3] + + Mask = Mask.to(device) + + # Distance matrix helper function + # def distance_matrix(x, y): + # return torch.cdist(x, y, p=2) + + @torch.no_grad() # check + def cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu): + # weight_param_key_optimizable [nk] + # exp_sq_dists_key_nonkey_edges [ne] + device=weight_param_key_optimizable.device + nk=weight_param_key_optimizable.shape[0] + + weight_param_key_optimizable_k=100*torch.abs(weight_param_key_optimizable) # [TODO]: 2000 # 100 + weight_param_edges=weight_param_key_optimizable_k[edges[:,1]] # (ne) + # assert all(weight_param_edges>0) + # assert all(sq_dists_key_nonkey_edges>0) + # weight_edges = (exp_sq_dists_key_nonkey_edges**(1/10))**weight_param_edges # TODO: to unify scale #test: 10 for block + weight_edges = exp_sq_dists_key_nonkey_edges**weight_param_edges # TODO: to unify scale #test: 20 for block + + weight_edges = torch.clamp(weight_edges, min=1e-10, max=1e10) # test + + weights_matrix=torch.zeros((nu,nk),device=device) + weights_matrix[edges[:,0],edges[:,1]]=weight_edges + sum_temp = torch.sum(weights_matrix,dim=-1,keepdim=True) + if any(sum_temp==0): + breakpoint() + if torch.isnan(weight_param_key_optimizable).any(): + breakpoint() + weights_matrix_normed=weights_matrix/sum_temp + + weight_edges_normed=weights_matrix_normed[edges[:,0],edges[:,1]] # (ne) + + return weight_edges_normed + + + + # prepare + Oris_xyzw_key_norm=Oris_xyzw_key/Oris_xyzw_key.norm(dim=-1,keepdim=True) + Oris_xyzw_key_optimized_norm=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) + n_neighbors=(edges[:,0]==0).sum() + if n_neighbors!=5: + breakpoint() + + # new + src_xyz_1=Transls[ts_most_confident,edges[::n_neighbors,0]] # (nu,3) + src_xyzw_1=Oris_xyzw[ts_most_confident,edges[::n_neighbors,0]] # (nu,4) + src_xyz_1 = nn.Parameter(src_xyz_1) + src_xyzw_1 = nn.Parameter(src_xyzw_1) + + # get sk_src_node and sk_dst_node (key nodes) + ts_most_confident_repeat=ts_most_confident.repeat(n_neighbors,1).T.reshape(-1) # (nt*5) + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + # optimizer = torch.optim.Adam( + # [weight_param_key_optimizable,src_xyz_1,src_xyzw_1,per_nonkeyGaussian_dweight], lr=lr + # ) + optimizer = torch.optim.Adam( + [weight_param_key_optimizable,src_xyz_1,src_xyzw_1], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, min_lr=min_lr + ) + ######################################################3 + + # ts_batches = torch.split(ts_all, batch_size) + # for ts_batch in ts_batches: + # Optimization loop + # for epoch in range(max_epochs): + # ts_all=torch.arange(nt_all,device=device) + # batch_size=30 + # pbar=tqdm(range(max_epochs)) + # for step in pbar: + # optimizer.zero_grad() + ts_all = torch.arange(nt_all, device=device) + batch_size = 30 + steps_per_epoch = (nt_all + batch_size - 1) // batch_size + total_steps = steps_per_epoch * max_epochs + + pbar = tqdm(range(total_steps), desc="Training") + + for step in pbar: + epoch = step // steps_per_epoch + inner_step = step % steps_per_epoch + + # Shuffle once per epoch + if inner_step == 0: + ts_shuffled = ts_all[torch.randperm(nt_all)] + + # Select current batch + start = inner_step * batch_size + ts_batch = ts_shuffled[start:start + batch_size] + pbar.set_description(f"Epoch {epoch}, Step {inner_step}/{steps_per_epoch-1}") + + optimizer.zero_grad() + + # print(f"ts_batch: {ts_batch}") + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # Mask_DQB=Mask[ts_batch] + + sk_nt_node_xyz_1=Transls_key[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_2=sk_nt_node_xyz_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_1=Oris_xyzw_key_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_1=sk_nt_node_xyzw_1[:,:,[3,0,1,2]] + sk_nt_node_quat_2=sk_nt_node_wxyz_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,4) + sk_nt_dst_node_xyz=sk_nt_node_xyz_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat=sk_nt_node_quat_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # error propagation + I0=torch.eye(3,device=device).expand(nt,nu,-1,-1).clone() # [nt,nu,3,3] + # I0[...,-1,-1]=Depths_info_perG_nonkey # 0.001 camel # CHECK TODO + # I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nu,3,3] + I1=I0 # test # [nt,nu,3,3] + I1_masked=I1[Mask_DQB] # [n_mask,3,3] + + if torch.isnan(weight_param_key_optimizable).any(): + print("weight_param_key_optimizable is nan") + breakpoint() + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + src_xyz=src_xyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask<=nt*nu,4) + + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) # [nu,n_neighbors] + # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # sk_w=weight_per_nonkeyGaussian_adjusted.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz, # [N,K,3] + sk_nt_dst_node_quat, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + # Transls_interpolation=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + # Oris_interpolation=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_interpolation_masked=mu_dst # [n_mask,3] + Oris_interpolation_masked=quat_dst_xyzw # [n_mask,4] + + + # error propagation + # a1=(Transls_interpolation[Mask]-Transls[Mask])[:,None,:] # (n_mask, 1, 3) + a1=(Transls_interpolation_masked-Transls[ts_batch][Mask_DQB])[:,None,:] # (n_mask, 1, 3) + # a1=(Transls_interpolation_masked-Transls[ts_batch])[:,None,:] # (n_mask, 1, 3) # ablation of loss + a1_T=a1.transpose(-1, -2) # (n_mask, 3, 1) + squared_norm=(a1@I1_masked@a1_T).squeeze(-1,-2) # [n_mask] + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # [n_mask] # add epsilon for numerical stability + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) # [nt,nu] + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) + # Dist_confident_pts_error=torch.norm(Transls_interpolation[Mask]-Transls[Mask],dim=-1) # [nt,nu] # last one + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Ori_confident_pts_error=torch.norm(Oris_interpolation[Mask]-Oris_xyzw[Mask],dim=-1) # [n_mask] + Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] + # Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] # ablation of loss + loss_ori_confident_pts = torch.mean(Ori_confident_pts_error) + + # Loss 3: Regularization on weight distance + # loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_velocity = torch.mean(torch.abs(Transls_interpolation[:-1]-Transls_interpolation[1:])) # (nt-1,nk,nk) # [TODO] + # loss_ori_velocity = torch.mean(torch.abs(Oris_interpolation[:-1] - Oris_interpolation[1:])) + + # loss_accel = compute_accel_loss(Transls_interpolation) + # loss_ori_accel=compute_accel_loss(Oris_interpolation) + + + # Loss 6: regularize per nonkey Gaussian delta weight + # loss_per_nonkeyGaussian=torch.mean(torch.abs(per_nonkeyGaussian_dweight)) + # loss_per_nonkeyGaussian = 1000*torch.mean(per_nonkeyGaussian_dweight ** 2) # TODO + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + a_loss_ori_confident_pts=1*loss_ori_confident_pts + # a_loss_interpolation=1*loss_interpolation # 100 + # a_loss_interpolation_ori=1*loss_interpolation_ori # 100 + # a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point + + # a_loss_velocity=0.1*loss_velocity + # a_loss_ori_velocity=0.1*loss_ori_velocity + # a_loss_accel=0.1*loss_accel + # a_loss_ori_accel=0.1*loss_ori_accel + + # a_loss_per_nonkeyGaussian=1e-2*loss_per_nonkeyGaussian + + # a_loss_sparsity=1e-3*loss_sparsity + + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + + total_loss = a_loss_dist_confident_pts + a_loss_ori_confident_pts #+ a_loss_per_nonkeyGaussian#+ \ + # a_loss_velocity + a_loss_ori_velocity + a_loss_accel + a_loss_ori_accel #+ a_loss_sparsity \ + # a_loss_dist + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness+ a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # a_loss_rigidity + # a_loss_interpolation + a_loss_interpolation_ori + \ + + # Backpropagation and optimization step + total_loss.backward() + # total_loss.backward(retain_graph=True) + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # # Logging progress + if epoch % log_interval == 0 or epoch == max_epochs - 1: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + f"a_loss_ori_confident_pts: {a_loss_ori_confident_pts.item():.6f}, " + # f"a_loss_per_nonkeyGaussian: {a_loss_per_nonkeyGaussian.item():.6f}, " + # f"a_loss_interpolation: {a_loss_interpolation.item():.6f}, " + # f"a_loss_interpolation_ori: {a_loss_interpolation_ori.item():.6f}, " + # f"a_loss_rigidity: {a_loss_rigidity.item():.6f}, " + # f"a_loss_velocity: {a_loss_velocity.item():.6f}, " + # f"a_loss_ori_velocity: {a_loss_ori_velocity.item():.6f}, " + # f"a_loss_accel: {a_loss_accel.item():.6f}, " + # f"a_loss_ori_accel: {a_loss_ori_accel.item():.6f}, " + # f"Density L: {a_loss_density.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + # # print(f"weight_normed: {(weight_normed>1e-6).sum().item():.6f}") + # # print(f"weight_param_key_optimizable:" {weight_param_key_optimizable}) + + pbar.set_description(f"L:{total_loss.item():.6f}, "+ + f"c:{a_loss_dist_confident_pts.item():.6f}, "+ + f"oc:{a_loss_ori_confident_pts.item():.6f}, "+ + # f"per:{a_loss_per_nonkeyGaussian.item():.6f}, "+ + # f"i: {a_loss_interpolation.item():.6f}, "+ + # f"io: {a_loss_interpolation_ori.item():.6f}, "+ + # f"r:{a_loss_rigidity.item():.6f}, "+ + # f"b:{a_loss_velocity.item():.6f}, "+ + # f"ov:{a_loss_ori_velocity.item():.6f}, "+ + # f"a:{a_loss_accel.item():.6f}, "+ + # f"oa:{a_loss_ori_accel.item():.6f}, "+ + f"lr:{optimizer.param_groups[0]['lr']:.2e}") + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + # # TEMP: for keeping data structure + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + # sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) + # # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # # sk_w=weight_per_nonkeyGaussian_adjusted[None,:,:].repeat(nt,1,1)[Mask_DQB] # (n_mask,n_neighbors) + + # weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + # weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + # # for i,e in enumerate(edges): + # # weight_distance_optimizable[e[0],e[1]]=weight_edges[i] + # weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + # print("End of optimize_others_with_orientation_edges4.") + # # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + # print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + # blend_dict = { + # 'sk_w': sk_w, + # 'src_xyz': src_xyz, + # 'src_quat': src_quat, + # 'sk_nt_src_node_xyz': sk_nt_src_node_xyz, + # 'sk_nt_src_node_quat': sk_nt_src_node_quat, + # 'sk_nt_dst_node_xyz': sk_nt_dst_node_xyz, + # 'sk_nt_dst_node_quat': sk_nt_dst_node_quat + # } + + + + # ********************* new trying begin ******************* + # Step 1: calculate src_xyz_1_new and src_xyzw_1_new + weight_param_key_optimizable=weight_param_key_optimizable.abs() + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # (nu,n_neighbors) + + weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + + # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + sk_w=weight_per_nonkeyGaussian # (nu,n_neighbors) + + sk_dst_node_xyz_1=Transls_key_optimized[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_dst_node_xyzw_1=Oris_xyzw_key_optimized_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_dst_node_wxyz_1=sk_dst_node_xyzw_1[:,[3,0,1,2]] + sk_dst_node_xyz_2=sk_dst_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_dst_node_quat_2=sk_dst_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + # breakpoint() + mu_dst, quat_dst=__DQB_warp2__( # [nu,3], [nu,4] + sk_w, # [N,K] + src_xyz_1, # [N,3] + sk_src_node_xyz_2, # [nu,n_neighbors,3] + sk_src_node_quat_2, # [nu,n_neighbors,4] + sk_dst_node_xyz_2, # [N,K,3] + sk_dst_node_quat_2, # [N,K,4] + dyn_o=None, + src_quat=src_wxyz_1, # [N,4] # wxyz + ) + src_xyz_1_new=mu_dst + src_xyzw_1_new=quat_dst[:,[1,2,3,0]] # [nu,4] + # breakpoint() + # for further optimization + blend_opt_dict = { + 'weight_param_key_optimizable': weight_param_key_optimizable, # (nu,n_neighbors) + 'src_xyz_1': src_xyz_1_new, # (nu,3) # CHECK + 'src_xyzw_1': src_xyzw_1_new, # (nu,4) # CHECK + 'per_nonkeyGaussian_dweight': per_nonkeyGaussian_dweight, # (nu,n_neighbors) + 'exp_sq_dists_key_nonkey_edges': exp_sq_dists_key_nonkey_edges, # (ne) + 'ts_most_confident': ts_most_confident, # (nu) + } + + # Step 2: calculate Transls_optimized and Oris_xyzw_optimized + Transls_optimized,Oris_xyzw_optimized=[],[] + ts_batches = torch.split(ts_all, batch_size) + for ts_batch in ts_batches: + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,4) + + # key points dst needed to be updated; src is the same + sk_nt_node_xyz_op_1=Transls_key_optimized[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_op_2=sk_nt_node_xyz_op_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_op_1=Oris_xyzw_key_optimized_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_op_1=sk_nt_node_xyzw_op_1[:,:,[3,0,1,2]] + sk_nt_node_quat_op_2=sk_nt_node_wxyz_op_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_dst_node_xyz_op=sk_nt_node_xyz_op_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat_op=sk_nt_node_quat_op_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # nonkey points update when src_xyz_1 and src_xyzw_1 are trainable + src_xyz=src_xyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask<=nt*nu,4) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz_op, # [N,K,3] + sk_nt_dst_node_quat_op, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + Transls_optimized_batch=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + Oris_xyzw_optimized_batch=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_optimized.append(Transls_optimized_batch) + Oris_xyzw_optimized.append(Oris_xyzw_optimized_batch) + Transls_optimized=torch.cat(Transls_optimized,dim=0) # [nt_all,nu,3] + Oris_xyzw_optimized=torch.cat(Oris_xyzw_optimized,dim=0) # [nt_all,nu,4] + + # ********************* new trying end ******************* + # breakpoint() + blend_dict={} + return Transls_optimized, Oris_xyzw_optimized, weight_param_key_optimizable,weight_distance_optimizable,blend_dict,blend_opt_dict + +def get_knn_edges_key(Transls,Mask,k): + # group by variance of relative motion + # Transls [nt,nk,3] + # std_threshold=0.01 + # max_dist_threshold=0.4 + + def nanstd(o, dim, keepdim=False): + result = torch.sqrt( + torch.nanmean( + torch.pow( torch.abs(o-torch.nanmean(o,dim=dim).unsqueeze(dim)),2), + dim=dim + ) + ) + if keepdim: + result = result.unsqueeze(dim) + return result + + nu=Transls.shape[1] + + # Dist=torch.cdist(Transls, Transls, p=2) + Dist = torch.norm(Transls[:, :, None, :] - Transls[:, None, :, :], dim=-1) # [nt, nu, nk] + Mask_dist=Mask[:, :, None] * Mask[:, None, :] # [nt, nu, nk] + Dist_masked = torch.where(Mask_dist, Dist, torch.tensor(float('nan'), device=Dist.device)) + # std_dist=Dist.std(dim=0) # [nu,nk] + std_dist=nanstd(Dist_masked,dim=0) # [nu,nk] + + mean_dist_masked=Dist_masked.nanmean(dim=0) # [nu,nk] # may include nan + # mean_dist=Dist.mean(dim=0) # [nu,nk] + Dist_masked2 = torch.where(Mask.unsqueeze(1).expand(-1, nu, -1), Dist, torch.tensor(float('nan'), device=Dist.device)) + mean_dist_masked2=Dist_masked2.nanmean(dim=0) # [nu,nk] # should not include nan # may still have nan + # mean_dist_masked2=Dist.mean(dim=0) # [nu,nk] # overwrite [TEST] + # breakpoint() + assert torch.any(mean_dist_masked2.isnan())==False + mean_dist_masked = torch.where(~torch.isnan(std_dist), mean_dist_masked, mean_dist_masked2) + # std_dist[~torch.isnan(std_dist)].mean(),mean_dist[~torch.isnan(mean_dist)].mean() + Dist_masked2 = torch.where(Mask_dist, Dist_masked, torch.tensor(float('-inf'), device=std_dist.device)) + # np.histogram(std_dist.numpy(),bins=50) + # np.histogram(std_dist[~torch.isnan(std_dist)].cpu().numpy(), bins=50) + # std_dist.min() + + max_dist = torch.max(Dist_masked2, dim=0).values # has -inf + max_dist2 = torch.where(~torch.isinf(max_dist), max_dist, torch.tensor(float('inf'), device=max_dist.device)) + # np.histogram(max_dist2.numpy()) + # np.histogram(max_dist[~torch.isinf(max_dist)].cpu().numpy(), bins=10) + + # (std_dist<0.01).sum(),(max_dist[~torch.isinf(max_dist)]<0.4).sum(),(max_dist2<0.4).sum(),torch.isnan(std_dist).sum() + + # try 1: + # connectivity_matrix=(std_dist40: + conditiona=Contribs2>0.5*np.max(Contribs2,axis=0) # 0.707 # TODO + # conditionb=Contribs2>contrib_threshold + # condition2=np.logical_or(conditiona,conditionb) + condition2=conditiona + else: + conditionb=Contribs2>contrib_threshold + condition2=conditionb + Mask2=torch.tensor(condition2,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable2=Oris_xyzw2.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edges + # 2.1 position edges + Dists_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True) + + # 2.2 orientation edges + Ori_distances_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True)*2 + + # 3. edge weight + # 3.1 position stds + weight_distance_optimizable2=torch.zeros((nu,nk),dtype=torch.float32,requires_grad=True) + + # knn graph (max contribs version) + + # def o3d_knn_p(pts2,pts, num_knn): + # indices = [] + # sq_dists = [] + # pcd = o3d.geometry.PointCloud() + # pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + # pcd_tree = o3d.geometry.KDTreeFlann(pcd) + # for p in pts2: + # [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + # indices.append(i[0:]) + # sq_dists.append(d[0:]) + # return np.array(sq_dists), np.array(indices) + + + # i_s_knn_others=[] + # sq_dists_5=[] + # n_neighbors=5 # 5 # 10 can not work + # ts_most_confident=[] + # for i in range(nu): + # t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + # eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + # pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + # i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + # pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + # pts2=[pt2] # just one point + # sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors) + # sq_dists,indices= sq_dists[0],indices[0] + + # flag_use_key_point_key_neighbors=True + # if flag_use_key_point_key_neighbors: + # ## given a key point, get the knn n_neighbors key points + # indice_nearest_key=indices[0] # nearest key point index + # pt3=pts[indice_nearest_key] # nearest key point + # pts3=[pt3] # just one key point + # sq_dists, indices = o3d_knn_p(pts3, pts, n_neighbors) # inlcude itself + # sq_dists,indices= sq_dists[0],indices[0] + # ## + # # calcuate the distance between the i-th other point and the neighboring key points + # sq_dists=np.linalg.norm(pts[indices]-pt2,axis=-1) # [n_neighbors] + + + # i_s_knn=i_s[indices] # usage: ids[i_s_knn] + # i_s_knn_others.append(i_s_knn) + # sq_dists_5.append(sq_dists) + # ts_most_confident.append(t_most_confident) + # i_s_knn_others=torch.stack(i_s_knn_others) + # sq_dists_5=np.array(sq_dists_5) + # sq_dists_5=torch.tensor(sq_dists_5,dtype=torch.float32) + # ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # # edges_others.shape # nu*5 + + # knn graph (minmax distance + contrib version) + n_neighbors=5 + edges_others,mean_dists=get_knn_edges_others(Transls,Mask,Transls2,Mask2,k=n_neighbors) + sq_dists_5=mean_dists[edges_others[:,0],edges_others[:,1]].reshape(-1,n_neighbors) # [nu,n_neighbors] + # edges_others,dists_edges=get_knn_edges_naive(n_neighbors,Transls,Transls2,flag_remove_self=False) # ablation of uknn + # sq_dists_5=dists_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + ts_most_confident=np.argmax(Contribs2[:,:],axis=0) # [nu] + ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + torch.cuda.empty_cache() + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + + # Method 3: + # use Dual Quat Blending (refer to MoSca) + # sq_dists_key_nonkey=-torch.ones([nu,nk]) + # sq_dists_key_nonkey[edges_others[:,0],edges_others[:,1]]=sq_dists_5.reshape(-1) + sq_dists_key_nonkey_edges=sq_dists_5.reshape(-1) # [ne] + weight_param_key_optimizable=1*torch.ones(nk,dtype=torch.float32,requires_grad=True) # 2000 [TODO] + per_nonkeyGaussian_dweight=torch.zeros(nu,n_neighbors,dtype=torch.float32,requires_grad=True) + + # batch training along t + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + + Transls_optimized2,Oris_xyzw_optimized2, weight_param_key_optimized2,weight_distance_optimized2,blend_dict,blend_opt_dict=\ + optimize_others_with_orientation_edges4(\ + edges_others,\ + Transls2, Transls_optimizable2.clone(), Mask2, \ + Oris_xyzw2, Oris_xyzw_optimizable2.clone(), \ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + weight_param_key_optimizable,sq_dists_key_nonkey_edges,\ + ts_most_confident,per_nonkeyGaussian_dweight,\ + W2Cs,Depths_info_perG_nonkey, + device) + + + # keep + edges_others_optimized=edges_others + Dists_optimized2=Dists_optimizable2 + Ori_distances_optimized2=Ori_distances_optimizable2 + + return Transls_optimized2,Dists_optimized2,weight_distance_optimized2,Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others,weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict + + + + + + +def main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges,threshold_min_set): + import sys + import os + sys.path.append(os.path.dirname("../flow3d")) + + import torch + import numpy as np + from scipy.spatial.transform import Rotation + import matplotlib.pyplot as plt + import pickle + import random + import glob + + import pypose as pp + from torch import nn + + from itertools import chain + from scipy.spatial import KDTree,Delaunay,distance_matrix + import open3d as o3d + from collections import defaultdict + import roma + + # set args + + # work_dir="/data/dataset/used_ds/som/sriracha-tree/work-dir" # backpack,haru-sit, sriracha-tree + # work_dir="/data/dataset/used_ds/som/backpack/work-dir" # work-dir1 # initialization results (no SoM model) + # work_dir="/data/dataset/custom_ds_trained/rhino" + # work_dir="/data/dataset/custom_ds_graph_model/rhino" + work_dir=save_dir + + # file_effective="effective_dict.pkl" # backpack 854 + # file_effective="effective_dict_949.pkl" # haru-sit + # file_effective="effective_dict_1908.pkl" # sriracha-tree # 1908 # 1000 + # file_effective="effective_dict_484.pkl" # backpack # work-dir1 + # file_effective="effective_dict_712.pkl" # rhino + + # dict_dir=work_dir+"/out_dicts2" # has w2c and camera's K + dict_dir=work_dir+"/out_dicts3" # has w2c and camera's K + + flag_draw_contribs_sample=False + flag_visulize_sparisity=False + flag_visulize_key_point=False + flag_visulize_nonkey_point=False + + flag_auto_threshold=False + contrib_threshold=2 # usually 2 + threshold=2e-0 ## TODO # for non-key points # fixed threshold + # threshold=-torch.inf + + + + # [1] load data + nt=count_out_dict_files(dict_dir) + dicts=[] + for t in range(nt): + with open(f"{dict_dir}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # print(dicts[0].keys()) + + with open(f"{dict_dir}/{file_effective}", 'rb') as file: + effective_dict = pickle.load(file) + inds_effective = effective_dict['ids_eff2'] #ids_eff1 + # Contribs_up = effective_dict['Contribs_up'] + threshold_min = effective_dict['threshold_min'] + threshold_mean = effective_dict['threshold_mean'] + Contribs_all = effective_dict['Contribs'] + print(f"threshold_min:{threshold_min}, threshold_mean:{threshold_mean}") + + # threshold_min=np.max([threshold_min-1, 0.5]) # 0.5 + # threshold_min=np.max([threshold_min-3, 0.5]) # 0.5 # paper-windmill test + # threshold_min=0.1 # test: believe nearly all points # paper-windmill + # threshold_min=0.5 # test: camel,backpack + if threshold_min_set!=-1: + print(f"set threshold_min to {threshold_min_set}.") + threshold_min=threshold_min_set + else: + print(f"use threshold_min from effective_dict: {threshold_min}.") + + # breakpoint() + if flag_auto_threshold: + print(f"set auto threshold to {threshold_min}.") + contrib_threshold=threshold_min + threshold=threshold_min + # contrib_threshold=threshold_mean + # threshold=threshold_mean + + print(f"set threshold_min to {threshold_min}.") + + device='cuda' + ids=inds_effective + # ids=np.array(inds_effective)[np.arange(0,200,6)].tolist() # quick test + # ids=np.array(inds_effective)[np.arange(0,854,2)].tolist() + + i_target=0 + i_show=10 + Transls=np.array([dicts[t]["means"][ids].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz=np.array([dicts[t]["quats"][ids].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw=Quats_wxyz[...,[1,2,3,0]] # xyzw + # Contribs=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) #### + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids] for t in range(nt)]) #### + Contribs=Contribs_all[:,ids] + # Contribs=Contribs_up[:,ids] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort=np.array([get_order(Contribs[:,i]) for i in range(Contribs.shape[1])]).transpose(1,0) + # Vars=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) + W2Cs=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + SE3s=np.concatenate([Transls,Quats_xyzw],axis=-1) # xyz xyzw (required by pypose) + SE3s=torch.tensor(SE3s,dtype=torch.float32) #,device='cuda') + Oris_xyzw=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type) + Oris=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type).Log() + lie_tensor_SE3s=pp.LieTensor(SE3s, ltype=pp.SE3_type) + Depths_info_perG=torch.stack([dicts[t]['depths_info_perG'].cpu() for t in range(nt)]) + # depth_ratio=0.001 # 0.001 + Depths_info_perG=depth_ratio*torch.ones_like(Depths_info_perG) # TEST + print(f"TEST: set Depths_info_perG={depth_ratio}*torch.ones_like(Depths_info_perG).") # test + Depths_info_perG_key=Depths_info_perG[:,ids] + # len(dicts) + # vars=Vars[:,i_target] + nk=len(ids) # num of key points + # np.histogram(vars[vars<1e0]) + print("nk:",nk) + print("dicts[0]['means'].shape:", dicts[0]["means"].shape) + + # draw + if flag_draw_contribs_sample: + # i_show=20 + # means=Transls[:,i_show] + # vars0=Vars[:,i_show] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars0,axis=1)),cmap='viridis') + + # means=Transls[:,i_target] + # vars1=Vars[:,i_target] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars1,axis=1)),cmap='viridis') + + plt.figure(figsize=(10,4)) + # plt.subplot(1,3,1) + # plt.plot(vars0,c='r') + # plt.plot(vars1,c='b',alpha=0.6) + # plt.xlabel("time frame") + # plt.ylabel("uncertainty") + # plt.ylim([0,1]) + i_show=123 + plt.subplot(1,3,2) + plt.plot(Contribs[:,i_show],c='r') + # plt.plot(Contribs[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution") + plt.xlabel("time frame") + plt.subplot(1,3,3) + plt.plot(Contribs_sort[:,i_show],c='r') + # plt.plot(Contribs_sort[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution sort") + plt.xlabel("time frame") + plt.show() + + + # [2] Graph initialization for key Gaussians + # Initializations + # 0. naive ori distance + Oris_rel=calculate_edge_relative_orientation(Oris_xyzw) + Ori_distances_naive=get_navie_ori_distances(Oris_rel) # (nk, nk) + + # 1. vertices + # 1.1 positions + Transls=torch.tensor(Transls,dtype=torch.float32) + Transls_optimizable=Transls.clone().requires_grad_(True) + # Transls_optimizable=torch.tensor(Transls,dtype=torch.float64,requires_grad=True) + # Transls_optimizable=torch.nn.Parameter(Transls_optimizable) + # condition1a=Contribs>0.4*np.max(Contribs,axis=0) # 0.707 + # # condition1b=Contribs2>contrib_threshold + # # condition1=np.logical_or(condition1a,condition1b) + # condition1=condition1a + # Mask=torch.tensor(condition1,dtype=torch.bool) + Mask=torch.tensor(Contribs>contrib_threshold,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable=Oris_xyzw.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edge distance + Dists = np.array([distance_matrix(Transls[t], Transls[t]) for t in range(nt)]) + Dists_optimizable=np.mean(Dists,axis=0) + Dists_optimizable=torch.tensor(Dists_optimizable,dtype=torch.float32,requires_grad=True) + + Ori_distances_optimizable=Ori_distances_naive.clone() + Ori_distances_optimizable=torch.tensor(Ori_distances_optimizable,dtype=torch.float32,requires_grad=True) + + # 3. edge weight + weight_distance_optimizable=1*torch.ones(Dists_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_distance_optimizable=weight_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # use the different weight + weight_ori_distance_optimizable=1*torch.ones(Ori_distances_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_ori_distance_optimizable=weight_ori_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # 4. get local edges + n_neighbors=120 + if version_key_edges=="max_contrib": + edges_graph=get_knn_edges(n_neighbors,Transls,Contribs,Transls,Contribs,contrib_threshold,flag_remove_self=True) # edge version: max contrib version + elif version_key_edges=="minmax_distance_contrib": + # edges_graph,_=get_knn_edges_others(Transls,Mask,Transls,Mask,k=n_neighbors) # edge version: minmax distance + contrib mask version + edges_graph,_=get_knn_edges_key(Transls,Mask,k=n_neighbors) + elif version_key_edges=="max_native": + edges_graph,_=get_knn_edges_naive(n_neighbors,Transls,Transls,flag_remove_self=True) # edge version: max contrib version + # breakpoint() + else: + breakpoint() + raise ValueError("version_key_edges should be max_contrib or minmax_distance_contrib or max_native") + # edges_graph.shape # [12395, 2] + weight_distance_optimizable=torch.zeros(Dists_optimizable.shape,dtype=torch.float32) + weight_distance_optimizable[edges_graph[:,0],edges_graph[:,1]]=1 #1e-5 + weight_distance_optimizable.requires_grad=True + + # [3] optimization for key Guassians + print("===========================================") + print("Start Optimization for key Guassians") + print("-------------------------------------------") + + torch.cuda.empty_cache() + Transls_optimized,Dists_optimized,weight_distance_optimized,Oris_xyzw_optimized, Ori_distances_optimized=\ + optimize_overall_with_orientation_edges(\ + edges_graph,\ + Transls, Transls_optimizable.clone(), Mask, Dists_optimizable.clone(), weight_distance_optimizable.clone(),\ + Oris_xyzw, Oris_xyzw_optimizable.clone(), Ori_distances_optimizable.clone(), weight_ori_distance_optimizable, \ + None, None, None,None,\ + W2Cs,Depths_info_perG_key) + + print("-------------------------------------------") + print("Finish Optimization for key Guassians") + print("===========================================") + + if flag_visulize_sparisity: + # breakpoint() + weight_distance_optimized[2,:10],weight_distance_optimized[:10,2] # check symetry + # look at normed weight + torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None] + data = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=50) # check dist + # np.sort(data[4])[-100:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + # (data>1e-6).sum() + np.histogram((data>1e-6).sum(axis=-1)) + # ((data>1e-6).sum(axis=-1)==3).sum() + + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + # Example 2D array + print((data>0.02).sum()) + + # Create a meshgrid for the X and Y coordinates + x = np.arange(data.shape[1]) + y = np.arange(data.shape[0]) + x, y = np.meshgrid(x, y) + + # Create a figure and a 3D axis + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the surface + ax.plot_surface(x, y, np.flip(data,axis=1), cmap='viridis') + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_yticks([0,100,200,300,400]) + ax.set_yticklabels(nk-np.arange(0,500,100)) + ax.set_zlabel('Weight Values') + # ax.set_zlim(0,0.5) + ax.set_title('Edge Weights') + + plt.show() + + + if flag_visulize_key_point: + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=Transls.shape[0]-1 # 0 35 80 100 150 + # t=33 # test backpack + # t=0 # 306 block + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=2) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=2) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + # save + # get optimized graph from normed weights + norm_weight_distance_optimized = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() + edges_graph_optimized=[] + for i in range(nk): + for j in range(nk): + if norm_weight_distance_optimized[i,j]>1e-6: + edges_graph_optimized.append([j,i]) + edges_graph_optimized=torch.tensor(edges_graph_optimized,dtype=torch.long) + + Ori_opt=Oris_xyzw_optimized + Nodes_opt=torch.concatenate([Transls_optimized.transpose(1,0),Ori_opt.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep=[[] for i in range(nk)] + # for e in edges_opt: + for e in edges_graph_optimized: + # Edges_keep[e[0]].append([e[1],e[0]]) + Edges_keep[e[1]].append([e[0],e[1]]) + + Edges_keep=[torch.tensor(Edges_keep[i],device=device) for i in range(nk)] + num_Edges_keep=[Edges_keep[i].shape for i in range(nk)] + # num_Edges_keep + + # check + torch.norm(Oris_xyzw_optimized[0],dim=-1)[:10] + + # for save + optimized_key_output={"Transls_optimized":Transls_optimized,\ + "Dists_optimized":Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized} + + # Save variables to a file + # torch.save({ + # 'dicts': dicts,\ + # 'ids': ids,\ + # 'Contribs': Contribs,\ + # 'Transls': Transls,\ + # 'Oris_xyzw': Oris_xyzw,\ + # 'Transls_optimizable': Transls_optimizable,\ + # 'Dists_optimizable': Dists_optimizable,\ + # 'weight_distance_optimizable':weight_distance_optimizable,\ + # 'Ori_distances_optimizable':Ori_distances_optimizable,\ + # 'Transls_optimized': Transls_optimized,\ + # 'Dists_optimized': Dists_optimized,\ + # "weight_distance_optimized":weight_distance_optimized,\ + # "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + # "Ori_distances_optimized":Ori_distances_optimized,\ + # "edges_graph":edges_graph,\ + # "edges_graph_optimized":edges_graph_optimized,\ + # "Nodes_opt":Nodes_opt,\ + # "Edges_keep":Edges_keep,\ + # "optimized_key_output":optimized_key_output,\ + # # "Transls_optimized1":Transls_optimized1,\ + # # "Dists_optimized1":Dists_optimized1,\ + # # "weight_distance_optimized1":weight_distance_optimized1,\ + # "device":device,\ + # "work_dir":work_dir}\ + + # , 'variables_8.pth') + + flag_read=False + if flag_read: + # Read variables from the file + checkpoint = torch.load('variables_8.pth') + dicts = checkpoint['dicts'] + ids = checkpoint['ids'] + Contribs = checkpoint['Contribs'] + Transls = checkpoint['Transls'] + Oris_xyzw = checkpoint['Oris_xyzw'] + Transls_optimized = checkpoint['Transls_optimized'] + Dists_optimized = checkpoint['Dists_optimized'] + weight_distance_optimized = checkpoint['weight_distance_optimized'] + Oris_xyzw_optimized = checkpoint['Oris_xyzw_optimized'] + Ori_distances_optimized = checkpoint['Ori_distances_optimized'] + edges_graph = checkpoint['edges_graph'] + edges_graph_optimized=checkpoint['edges_graph_optimized'] + Nodes_opt = checkpoint['Nodes_opt'] + Edges_keep = checkpoint['Edges_keep'] + optimized_key_output = checkpoint['optimized_key_output'] + # Transls_optimized1 = checkpoint['Transls_optimized1'] + # Dists_optimized1 = checkpoint['Dists_optimized1'] + # weight_distance_optimized1 = checkpoint['weight_distance_optimized1'] + device=checkpoint['device'] + work_dir=checkpoint['work_dir'] + nk=len(ids) + + + # [4] Graph initialization for non-key Gaussians + # 1. prepare data for non-key Gaussians + Ng=dicts[0]["means"].shape[0] # num of all Gaussians + nt=len(dicts) + # Ng=1000 + ids2=[i for i in range(Ng) if i not in ids] # ids of non-target nodes + # Contribs2_temp=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2_temp=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2_temp=Contribs_all[:,ids2] + Contribs2_temp_max=Contribs2_temp.max(axis=0) + if nt>350: + n_nonkey_max=36000 # 36000 # 150000 72000 too large memory for block teddy wheel + elif nt>40: + n_nonkey_max=72000 + else: + n_nonkey_max=150000 + if Contribs2_temp_max.shape[0]>n_nonkey_max: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[-n_nonkey_max] # the nth (12000th/24000th) largest contrib value + else: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[0] + if nt>40: # iphone davis + contribs2_nth_largest=max(contribs2_nth_largest,0.01) # 0.1 + else: # nvidia + contribs2_nth_largest=max(contribs2_nth_largest,1e-4) + threshold=contribs2_nth_largest # rewrite threshold + ids2=np.array(ids2)[Contribs2_temp.max(axis=0)>threshold].tolist() # remove nodes with small contribs #15 + + + Transls2=np.array([dicts[t]["means"][ids2].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz2=np.array([dicts[t]["quats"][ids2].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw2=Quats_wxyz2[...,[1,2,3,0]] # xyzw + # Contribs2=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2=Contribs_all[:,ids2] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort2=np.array([get_order(Contribs2[:,i]) for i in range(Contribs2.shape[1])]).transpose(1,0) + # Vars2=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # W2Cs2=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + # SE3s2=np.concatenate([Transls2,Quats_xyzw2],axis=-1) # xyz xyzw (required by pypose) + # SE3s2=torch.tensor(SE3s2,dtype=torch.float32) #,device='cuda') + Oris_xyzw2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type) + Oris2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type).Log() + Depths_info_perG_nonkey=Depths_info_perG[:,ids2] + nu=len(ids2) # num of non-target nodes + print(f'Ng={Ng},nk=len(ids)={len(ids)},nu=len(ids2)={len(ids2)},threshold={threshold}') + + # # test threshold + # condition1=Contribs2>0.707*np.max(Contribs2,axis=0) + # condition2=Contribs2>contrib_threshold + # np.histogram((np.logical_or(condition1,condition1)).sum(axis=0),bins=30) + + # 2. run optimizaiton for non-key Guassians + print("===========================================") + print("Start Optimization for nonkey Guassians") + print("-------------------------------------------") + + Transls_optimized2,Dists_optimized2,weight_distance_optimized2,\ + Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others, \ + weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict \ + =run_batch(Transls2,Oris_xyzw2,Contribs2,\ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + Contribs,contrib_threshold,\ + W2Cs,Depths_info_perG_nonkey,\ + Mask) + + print("weight_param_key_optimized2 hist:",np.histogram(weight_param_key_optimized2.cpu().detach().numpy(),bins=10)) + + print("Finish Optimization for nonkey Guassians") + print("===========================================") + + + + if flag_visulize_nonkey_point: + # 3. check + # just for Method 1 + data = (torch.abs(weight_distance_optimized2)/torch.norm(weight_distance_optimized2,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=5) # check dist + np.sort(data[4])[-5:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + np.histogram((data>1e-6).sum(axis=-1)) # check sparsity + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=nt-1 # 0 35 80 100 150 + # t = 168 # backpack test + # t = 0 # 306 block # 33 + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + pts_others=Transls_optimized2.cpu().detach().numpy()[t] + # pts_others=Transls2.cpu().detach().numpy()[t] + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + x3 = pts_others[:,0] + y3 = pts_others[:,1] + z3 = pts_others[:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=4) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=4) + + ax.scatter(x3, y3, z3, c='k', label='other Points',s=1) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + Ori_opt2=torch.tensor(Quats_xyzw2,device=device) # tempory + # Ori_opt2=Oris_xyzw_optimized2.tensor() + Nodes_opt2=torch.concatenate([Transls_optimized2.transpose(1,0),Ori_opt2.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep2=[[] for i in range(nu)] + # for e in edges_opt: + for e in edges_others_optimized: + Edges_keep2[e[0]].append([e[1],e[0]+nk]) # [key id, other id + offset (nk)] + # Edges_keep2[e[1]].append([e[0],e[1]]) + + Edges_keep2=[torch.tensor(Edges_keep2[i],device=device) for i in range(nu)] + num_Edges_keep2=[Edges_keep2[i].shape for i in range(nu)] + + # combine key and others + Nodes_opt_combine=torch.cat([Nodes_opt,Nodes_opt2],dim=0) + Edges_keep_combine=Edges_keep+Edges_keep2 + ids_combine=ids+ids2 + + optimized_others_output={"Transls_optimized":Transls_optimized2,\ + "Dists_optimized":Dists_optimized2,\ + "weight_distance_optimized":weight_distance_optimized2,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized2, \ + "Ori_distances_optimized":Ori_distances_optimized2} + + + # [5] Save graph preprocessing data + quats_xyzw=Nodes_opt[:,:,3:] # xyzw + quats_wxyz=quats_xyzw[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt=[Nodes_opt[:,:,:3],quats_wxyz] # [[nk,nt,3], [nk,nt,4]] + + # combine + quats_xyzw_combine=Nodes_opt_combine[:,:,3:] # xyzw + quats_xyzw_combine=quats_xyzw_combine[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt_combine=[Nodes_opt_combine[:,:,:3],quats_xyzw_combine] # [[nk,nt,3], [nk,nt,4]] + + # tracks_SE3_opt[1].shape,Nodes_opt.shape,tracks_SE3_opt_combine[1].shape,Nodes_opt_combine.shape + + nk=len(Edges_keep) + adjacent_matrix=torch.zeros((nk,nk)) + for i in range(nk): + if Edges_keep[i].shape[0]==0: + continue + indices=Edges_keep[i][:,0] + adjacent_matrix[i,indices]=1 + + adjacent_matrix = torch.logical_or(adjacent_matrix.t(), adjacent_matrix).type(torch.int32) + + Edges_keep_bi=[] + for i in range(nk): + indices=torch.where(adjacent_matrix[i,:]==1)[0] + edges_keep_bi=torch.cat((indices.unsqueeze(0),i*torch.ones_like(indices).unsqueeze(0))).t() + Edges_keep_bi.append(edges_keep_bi) + + Edges_keep_bi_combine=Edges_keep_bi+Edges_keep2 + + + # Save 1: + is_target=list(range(len(ids))) + is_target_combine=list(range(len(ids_combine))) + opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + "optimized_key_output":optimized_key_output,\ + "Nodes_opt_combine":Nodes_opt_combine,"tracks_SE3_opt_combine":tracks_SE3_opt_combine,"ids_combine":np.array(ids_combine),\ + "is_target_combine":is_target_combine,"Edges_keep_combine":Edges_keep_combine,"Edges_keep_bi_combine":Edges_keep_bi_combine,\ + "optimized_others_output":optimized_others_output} + # opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + # "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + # "optimized_key_output":optimized_key_output} + import pickle + dict_dir=f"{work_dir}/opt_dicts" + os.makedirs(dict_dir, exist_ok=True) + file_name=f"opt_v8_{len(ids)}_0.pkl" + # file_name=f"opt_427_var0.002_cut0.002_2.pkl" # 854 eff file + # file_name=f"opt_854_var0.002_cut0.002_nearby.pkl" # 854 eff file + # file_name=f"opt_2544_var0.002_cut0.002.pkl" # 2544 eff file + # file_name=f"opt_2038_var0.002_cut0.002_test.pkl" # 2544 eff file + # file_name="opt_dicts_test3.pkl" + + # # save + # file=open(f'{dict_dir}/{file_name}', 'wb') + # pickle.dump(opt_dict, file) + # file.close() + # print("file_path:",f'{dict_dir}/{file_name}') + + # Save 2: + # Save variables to a file + dict_dir_graph=f"{work_dir}/graph_model/" + os.makedirs(dict_dir_graph, exist_ok=True) + # file_name_graph=f"v8_712_parameters0.pth" + file_path_graph=dict_dir_graph+file_name_graph + + torch.save({ + 'dicts': dicts,\ + 'ids': ids,\ + 'Contribs': Contribs,\ + 'Transls': Transls,\ + 'Oris_xyzw': Oris_xyzw,\ + 'Transls_optimizable': Transls_optimizable,\ + 'Dists_optimizable': Dists_optimizable,\ + 'weight_distance_optimizable':weight_distance_optimizable,\ + 'Ori_distances_optimizable':Ori_distances_optimizable,\ + 'Transls_optimized': Transls_optimized,\ + 'Dists_optimized': Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized,\ + "edges_graph": edges_graph,\ + "edges_graph_optimized": edges_graph_optimized,\ + "Nodes_opt":Nodes_opt,\ + "Edges_keep":Edges_keep,\ + "optimized_key_output":optimized_key_output,\ + # "Transls_optimized1":Transls_optimized1,\ + # "Dists_optimized1":Dists_optimized1,\ + # "weight_distance_optimized1":weight_distance_optimized1,\ + "device":device,\ + "work_dir":work_dir,\ + "ids2":ids2,\ + "Contribs2": Contribs2,\ + "Transls_optimized2":Transls_optimized2,\ + "Dists_optimized2":Dists_optimized2,\ + "weight_distance_optimized2":weight_distance_optimized2,\ + "Oris_xyzw_optimized2":Oris_xyzw_optimized2,\ + "Ori_distances_optimized2":Ori_distances_optimized2,\ + "edges_others_optimized":edges_others_optimized,\ + "edges_others":edges_others,\ + "blend_opt_dict":blend_opt_dict,\ + "contrib_threshold":contrib_threshold,\ + }\ + , file_path_graph) + + print("file_path_graph:",file_path_graph) + + +def main2(save_dir,voxel_size=0.4,n_min_effective_frames=5,ratio_key=-1): + flag_remove_outlier=False + flag_show_sample=False + flag_show_threshold=False + flag_ablation_select_pts_wo_uncertainty=False + flag_ablation_random_select=False + # ratio_key=0.08 # ablation study of threshold and selected number of pts at each step. + + + # folder_dict="/data/dataset/used_ds/som/sriracha-tree/work-dir/out_dicts2" # haru-sit, backpack, sriracha-tree + # folder_dict="/data/dataset/custom_ds_graph_model/rhino/out_dicts3" # rhino + folder_dict=f"{save_dir}/out_dicts3" + + nt=count_out_dict_files(folder_dict) + dicts=[] + for t in range(nt): + with open(f"{folder_dict}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy() for t in range(nt)]) #### (old) + Contribs=np.array([dicts[t]['contribs_strict'].cpu().numpy() for t in range(nt)]) #### + + # Safety check for empty Contribs + if Contribs.size == 0: + breakpoint() + print("Warning: Contribs array is empty, using contribs_in_mask as fallback") + + if ratio_key>0: + max_num_key_nodes=int(Contribs.shape[1]*ratio_key) + num_stage_wise_select_key_nodes=int(max_num_key_nodes/1.5) + max_num_key_nodes = min(max_num_key_nodes, 1500) # avoid used memory is too large. + max_num_key_nodes = max(max_num_key_nodes, 5) # avoid no of key nodes too few. + else: + max_num_key_nodes=1500 # preivous setting + num_stage_wise_select_key_nodes=1000 # previous setting + + print("******ratio_key: ",ratio_key, "max_num_key_nodes: ",max_num_key_nodes," num_stage_wise_select_key_nodes: ",num_stage_wise_select_key_nodes) + + + c99=np.quantile(Contribs, 0.99) + c80=np.quantile(Contribs, 0.80) + contrib_ref=Contribs[(Contribsc80)].mean() + if contrib_ref<0.5: + prod_num=0.5/contrib_ref + Contribs=Contribs*prod_num + Contribs = np.clip(Contribs, a_min=None, a_max=10) + # breakpoint() + # dicts[0].keys(),nt,dicts[0]['means'].shape + + if flag_remove_outlier: + def get_outliers_by_density(means): + X=means[np.arange(0,means.shape[0],1),:] + + # Apply HDBSCAN clustering + # clusterer = hdbscan.HDBSCAN(min_cluster_size=10, min_samples=5, allow_single_cluster=True) + # labels = clusterer.fit_predict(X) + clusterer = DBSCAN(eps=0.02, min_samples=10) ############### hyperparameter!!! + labels=clusterer.fit(X).labels_ + outlier_map=labels==-1 + + # Unique clusters + # n_clusters = len(set(labels)) - (1 if -1 in labels else 0) # Ignore noise label (-1) + # print(f"Detected Clusters: {n_clusters}") + # assert n_clusters==1 + + return outlier_map + + outlier_map=get_outliers_by_density(dicts[t]['means'].cpu().numpy()) + outlier_map.sum() + + Contribs_up=[] + for t in range(nt): + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + if flag_remove_outlier: + outlier_map=get_outliers_by_density(means) + contribs[outlier_map]=1e-10 ## + Contribs_up.append(contribs) + Contribs_up=np.array(Contribs_up) + else: + Contribs_up=None + + print("num of gaussians: ",dicts[0]['means'].shape[0]) + # 1st Gaussian selection round + # select according to contribs + selected_ids_ts=[] + for t in range(nt): + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + else: + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) # ascending order w.r.t. ids + n=contribs.shape[0] + # candidate_ids_map=sorted_indices>=n-1000 + candidates_in_raw_ids=sorted_indices[-num_stage_wise_select_key_nodes:] #1000 + # candidates_in_raw_ids=[i for i in range(n) if candidate_ids_map[i]] + + candidate_means=means[candidates_in_raw_ids] + # candidate_contribs=contribs[candidate_ids_map] + + # voxel_size = 0.4 # 0.5 # 0.4 + voxel_grid = np.floor(candidate_means / voxel_size) + _, unique_indices = np.unique(voxel_grid, axis=0, return_index=True) + selected_candidates_in_raw_ids=[candidates_in_raw_ids[i] for i in unique_indices] + if flag_ablation_random_select: + n_random_select=len(selected_candidates_in_raw_ids) + selected_candidates_in_raw_ids=np.random.choice(selected_candidates_in_raw_ids, n_random_select, replace=False).tolist() + selected_ids_ts.append(selected_candidates_in_raw_ids) + # print(f"t={t}, selected_candidates_in_raw_ids.shape={len(selected_candidates_in_raw_ids)}") + + counts=np.zeros(n) + for t in range(nt): + selected_candidates_in_raw_ids=selected_ids_ts[t] + counts[selected_candidates_in_raw_ids]+=1 + + selected_ids_effective_1frame=[i for i in range(n) if counts[i]>0] + print("#pts left after 1st selection round: ", len(selected_ids_effective_1frame)) + # counts[selected_ids_effective_1frame] + + # avoid too many point in the 1st selection round + # select pts + def select_pts(n,selected_ids_effective_1frame): + arr1=np.ones(n) + arr2=np.zeros(len(selected_ids_effective_1frame)-n) + arr=np.concatenate((arr1,arr2),axis=0) + np.random.shuffle(arr) + selected_ids_effective_1frame=np.array(selected_ids_effective_1frame)[arr.astype(bool)].tolist() + return selected_ids_effective_1frame + # if len(selected_ids_effective_1frame)>2200: + # selected_ids_effective_1frame=select_pts(2200,selected_ids_effective_1frame) + # print("#pts left after 1st selection round (adjusted): ", len(selected_ids_effective_1frame)) + + # 2nd Gaussian selection round + ## 2nd selection round: whose contribs are effective at least nt/6 frames + # n_min_effective_frames=30 # 30 + # n_min_effective_frames=int(nt/6) + # n_min_effective_frames=min(5,int(nt/6)) # use 30 to avoid key pts not find in the long video. # TODO # 5 for haru-sit + print("n_min_effective_frames: ",n_min_effective_frames) + counts_effective_frames=np.zeros(len(selected_ids_effective_1frame)) + thresholds=[] + for t in range(nt): + if t%100==0: + print(t) + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + else: + # means=dicts[t]['means'].cpu().numpy() + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) + # candidates_in_raw_ids=sorted_indices[-2000:] + threshold=contribs[sorted_indices[-num_stage_wise_select_key_nodes]] # the 2000th largest contrib # 1000 + threshold=max(threshold,0.5) # 0.5 + counts_effective_frames+=(contribs[selected_ids_effective_1frame]>=threshold).astype(int) + thresholds.append(threshold) + + threshold_min=np.min(np.array(thresholds)) + threshold_mean=np.mean(np.array(thresholds)) + print("min contrib threshold: ",threshold_min) + print("mean contrib threshold: ",threshold_mean) + print(np.histogram(counts_effective_frames)) + if flag_show_threshold: + import matplotlib.pyplot as plt + plt.figure() + plt.plot(thresholds) + plt.show() + selected_ids_eff=[selected_ids_effective_1frame[i] for i in range(len(selected_ids_effective_1frame)) if counts_effective_frames[i]>=n_min_effective_frames] + print("#pts left after 2nd selection round: ", len(selected_ids_eff)) + # counts_effective_frames[counts_effective_frames>30].shape + + # avoid too many point in the 2nd selection round + if len(selected_ids_eff)>max_num_key_nodes: # 1000 2500 # 1500 + selected_ids_eff=select_pts(max_num_key_nodes,selected_ids_eff) # 1000 2500 1500 + print("#pts left after 2nd selection round (adjusted): ", len(selected_ids_eff)) + + if flag_ablation_select_pts_wo_uncertainty: + selected_ids_eff=np.arange(Contribs.shape[1]) + selected_ids_eff=select_pts(1000,selected_ids_eff) # 1000 # 2500 1500 + print(f"flag_ablation_select_pts_wo_uncertainty={flag_ablation_select_pts_wo_uncertainty}") + print("random select key pts:", len(selected_ids_eff)) + + if flag_show_sample: + import matplotlib.pyplot as plt + # t=nt-1 + t=0 + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + x = means[selected_ids_eff][:,0] + y = means[selected_ids_eff][:,1] + z = means[selected_ids_eff][:,2] + # sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=np.log(contribs[selected_ids_eff]),cmap='viridis') + sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=contribs[selected_ids_eff],cmap='viridis') + + cbar = plt.colorbar(sc) + cbar.set_label('Contribs') + ax.set_xlabel('X Label') + ax.set_ylabel('Y Label') + ax.set_zlabel('Z Label') + ax.axis('equal') + plt.show() + + # save + file_path=f"{folder_dict}/effective_dict_{len(selected_ids_eff)}.pkl" + effective_dict={"ids_eff2":selected_ids_eff,"ids_eff1":selected_ids_effective_1frame,"Contribs_up":Contribs_up, + "threshold_min":threshold_min,"threshold_mean":threshold_mean,"Contribs":Contribs} + file=open(file_path, 'wb') + pickle.dump(effective_dict, file) + file.close() + print(file_path) + file_effective=f"effective_dict_{len(selected_ids_eff)}.pkl" + return file_effective + +def compute_knn_indices(points, k, chunk_size=1024): + """ + Memory-efficient KNN search: returns indices of k nearest neighbors for each point. + points: [N, 3] + returns: [N, k] LongTensor + """ + N = points.shape[0] + knn_indices = [] + + for i in range(0, N, chunk_size): + end = min(i + chunk_size, N) + chunk = points[i:end] # [chunk, 3] + dists = torch.cdist(chunk, points) # [chunk, N] + topk = dists.topk(k + 1, largest=False).indices[:, 1:] # exclude self + knn_indices.append(topk) + + return torch.cat(knn_indices, dim=0) # [N, k] + +def gaussian_kde_density_knn(points, inv_covariances, k=1000, eps=1e-6): + """ + Vectorized version using k nearest neighbors for KDE density. + points: [N, 3] + covariances: [N, 3, 3] + returns: [N] density using k-NN Mahalanobis distances + """ + N, D = points.shape + device = points.device + + # Invert covariances: [N, 3, 3] + # cov_inv = torch.inverse(covariances + eps * torch.eye(D, device=device)) + cov_inv = inv_covariances + + # # Get k-NN indices using Euclidean distance + # dists = torch.cdist(points, points) # [N, N] + # knn_indices = dists.topk(k + 1, largest=False).indices[:, 1:] # [N, k] (exclude self) + + # Memory-efficient KNN + chunk_size= 10000 + knn_indices = compute_knn_indices(points, k, chunk_size=chunk_size) # [N, k] + + # Gather neighbor points: [N, k, 3] + knn_points = points[knn_indices] # batched gather + + # Center point per row: [N, 1, 3] -> broadcast to [N, k, 3] + diffs = knn_points - points[:, None, :] # [N, k, 3] + + # Apply Mahalanobis: dᵢⱼ = (xᵢ - xⱼ)^T Σᵢ⁻¹ (xᵢ - xⱼ) + # cov_inv: [N, 3, 3], diffs: [N, k, 3] + # compute: diffᵢⱼ^T @ cov_invᵢ @ diffᵢⱼ -> result [N, k] + + diffs_unsq = diffs.unsqueeze(-1) # [N, k, 3, 1] + cov_inv_exp = cov_inv[:, None, :, :] # [N, 1, 3, 3] + m = torch.matmul(cov_inv_exp, diffs_unsq) # [N, k, 3, 1] + mahalanobis_sq = torch.matmul(diffs_unsq.transpose(-1, -2), m).squeeze(-1).squeeze(-1) # [N, k] + + # Gaussian kernel weights + weights = torch.exp(-0.5 * mahalanobis_sq) # [N, k] + + density = weights.sum(dim=1) # [N] + return density + + +def plot_density_histogram_rgb(density, bins=20, height=256): + """ + density: [N] tensor + height: desired output image height in pixels + Returns: [height, width, 3] RGB uint8 numpy image of the histogram with log-scale x-axis and labeled axes + """ + density_np = density.cpu().numpy() + density_np = density_np[density_np > 0] # filter out non-positive for log scale + + dpi = 100 # Dots per inch + fig_height_inches = height / dpi + aspect_ratio = 1.5 # width / height + fig_width_inches = fig_height_inches * aspect_ratio + + # Create figure with desired size + fig, ax = plt.subplots(figsize=(fig_width_inches, fig_height_inches), dpi=dpi) + ax.hist(density_np, bins=bins, color='skyblue', edgecolor='black') + ax.set_xscale("log") + ax.set_title("Density Histogram (Log X-axis)") + ax.set_xlabel("Density (log scale)") + ax.set_ylabel("Count") + + fig.tight_layout() + fig.canvas.draw() + + width, height = fig.canvas.get_width_height() + image = np.frombuffer(fig.canvas.tostring_rgb(), dtype=np.uint8).reshape(height, width, 3) + + plt.close(fig) + return image.astype(np.uint8) + +def load_gt_masks(ws, gt_mask_dirname="mask"): + import imageio.v2 as imageio + """Load ground truth masks from directory""" + gt_mask_dir = osp.join(ws, gt_mask_dirname) + if not osp.exists(gt_mask_dir): + logging.warning(f"GT mask directory {gt_mask_dir} not found") + breakpoint() + return None + + gt_mask_fns = [ + f for f in os.listdir(gt_mask_dir) + if f.endswith(".png") or f.endswith(".jpg") + ] + gt_mask_fns.sort() + + gt_masks = [] + for gt_mask_fn in gt_mask_fns: + gt_mask_path = osp.join(gt_mask_dir, gt_mask_fn) + gt_mask = imageio.imread(gt_mask_path) + if gt_mask.ndim == 3: + gt_mask = gt_mask[..., 0] # Take first channel if RGB + gt_mask = (gt_mask > 128).astype(np.float32) # Convert to binary + gt_masks.append(gt_mask) + + gt_masks = torch.Tensor(np.stack(gt_masks)).float() # T,H,W + # self.register_gradfree_buffer("gt_masks", gt_masks) + # self.has_gt_masks = True + logging.info(f"Loaded GT masks from {gt_mask_dir}: {gt_masks.shape}, foreground ratio: {gt_masks.mean():.3f}") + return gt_masks, gt_mask_fns + +# @torch.inference_mode() +def main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model=None): + import torch + import argparse + import os + import torch + import time + from leo_renderer import Renderer + from lib_render.render_helper import render, render_gsplat + import numpy as np + from tqdm import tqdm + import imageio.v3 as iio + + from datetime import datetime + from loguru import logger as guru + import torch.nn.functional as F + import cv2 + import matplotlib.pyplot as plt + import lpips + + from lib_ugraph.ugraph_utils import compute_ssim_map, compute_lpips_map + + # parser = argparse.ArgumentParser() + # parser.add_argument("--cfg_fn", type=str) + # parser.add_argument("--port", type=int, default=8899) + # parser.add_argument("--work_dir", type=str, required=True) + # args = parser.parse_args() + + + + if dataset_name=='iphone': + cfg_fn='/data/repo/MoSca/profile/iphone/iphone_fit.yaml' + elif dataset_name=='davis' or dataset_name=='davis_mask': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='nvidia': + cfg_fn='/data/repo/MoSca/profile/nvidia/nvidia_fit.yaml' + elif dataset_name=='diffusion4d': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='objaverse': + cfg_fn='/data/repo/MoSca/profile/objaverse/objaverse_fit.yaml' + else: + cfg_fn='' + breakpoint() + port=9000 + + device = torch.device("cuda") # if torch.cuda.is_available() else "cpu") + renderer = Renderer(cfg_fn, device, work_dir, port=port, bg_flag=True, fg_flag=True,load_s2d=True, ws=ws, use_ugraph=False, relative_dir_saved_mosca_model=relative_dir_saved_mosca_model) + + nt=renderer.cams.T + print("nt:",nt) + + W, H = renderer.cams.default_W, renderer.cams.default_H + assert W == renderer.s2d.W and H == renderer.s2d.H + K = renderer.cams.K(H=H,W=W) + # focal = 0.5 * H / np.tan(0.5 * camera_state.fov).item() + # K = torch.tensor( + # [[focal, 0.0, W / 2.0], [0.0, focal, H / 2.0], [0.0, 0.0, 1.0]], + # device=device, + # ) + # w2c = torch.linalg.inv( + # torch.from_numpy(camera_state.c2w.astype(np.float32)).to(self.device) + # ) + ts=torch.arange(nt,device=device) + w2cs=torch.stack([renderer.cams.T_cw(t) for t in ts],dim=0) + # with torch.inference_mode(): + # for w2c,t in tqdm(zip(w2cs,ts)): + # # w2c = renderer.cams.T_cw(t) + + # gs5 = [] + # if renderer.bg_flag: + # gs5.append(renderer.s_model()) + # if renderer.fg_flag: + # gs5.append(renderer.d_model(t)) + # render_dict, _ = render_gsplat( + # gs5, + # H, + # W, + # K=K, + # T_cw=w2c, + # bg_color=[1.0, 1.0, 1.0], + # ) + + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # # render_dict0 = render( + # # gs5, + # # H, + # # W, + # # K=K, + # # T_cw=w2c, + # # bg_color=[1.0, 1.0, 1.0], + # # ) + + # # img0 = torch.clamp(render_dict0["rgb"].permute(1, 2, 0), 0.0, 1.0) + # # img0 = (img0.cpu().numpy() * 255.0).astype(np.uint8) + # # combined = np.hstack([img0,img]) + # # iio.imwrite(f"./test/img_{t.item()}.png", combined) # check if they look the same. + # # compare_images(img, img0) + + if save_dir=='': + save_dir = work_dir + + video_dir = f"{save_dir}/videos/{datetime.now().strftime('%Y-%m-%d-%H%M%S')}" + print("*"*20) + print("*"*20) + print(f"Saving video to {video_dir}") + print("*"*20) + print("*"*20) + os.makedirs(video_dir, exist_ok=True) + + torch.backends.cuda.preferred_linalg_library("magma") + + # if dataset_name=="davis": + # prior_depths=torch.stack(train_dataset.depths).to(device) + # elif dataset_name=="iphone": + # prior_depths=train_dataset.depths.to(device) # [nt,H,W] + # else: + # # breakpoint() + # print("No prior depth!!!") + + if dataset_name=="davis" or dataset_name=="iphone" or dataset_name=="diffusion4d" or dataset_name=="objaverse" or dataset_name=="davis_mask": + thr_density_strict=0.1 #5 + thr_density_loose=0.1 #1 + thr_rgb_diff=0.3 + thr_ssim=0.01 + elif dataset_name=="nvidia": + thr_density_strict=0.0 + thr_density_loose=0.0 + thr_rgb_diff=0.6#0.6 + thr_ssim=0.001 # 0.001 + else: + breakpoint() + + try: + # gt_mask_dir = osp.join(ws, "mask") + # renderer.s2d.load_gt_masks(gt_mask_dir) + gt_masks=renderer.s2d.gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + except: + breakpoint() + # gt_masks=None + gt_masks, gt_mask_fns=load_gt_masks(ws, "mask") + gt_masks=gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + # breakpoint() + prior_depths=renderer.s2d.dep.to(device) # [nt,H,W] + # breakpoint() + prior_depths_grad, _, _=compute_depth_gradient(prior_depths) # [nt,H,W] + # depths_info = 0.05*torch.exp(-30*prior_depths_grad) # [nt,H,W] # TODO: hyperparameter + # depths_info = depths_info.clamp_min(1e-4) # set minimum value to 1e-4 + depths_info = 1e-3 * torch.ones_like(prior_depths_grad,device=device) # 1e-3 1e-5 + # depths_info = 1e-3 * torch.ones((nt,H,W),device=device) # 1e-3 1e-5 + # depths_info[prior_depths_grad <= 0.03] = 0.1 # [nt,H,W] # TODO: hyperparameter # 0.05 # TODO!!! + print("np.histogram:",np.histogram(prior_depths_grad[gt_masks].flatten().cpu().numpy(), bins=10)) + # np.histogram(depths_info.flatten().cpu().numpy(), bins=10) + depth_min = prior_depths.min().item() + def safe_quantile(tensor: torch.Tensor, q: float, max_samples: int = 100000): + """Compute quantile safely by subsampling if needed.""" + if tensor.numel() > max_samples: + indices = torch.randperm(tensor.numel(), device=tensor.device)[:max_samples] + tensor = tensor.view(-1)[indices] + return tensor.quantile(q) + depth_max = safe_quantile(prior_depths, 0.99).item() + prior_depths_grad_threshold = safe_quantile(prior_depths_grad[gt_masks], 0.90).item() + print(f"set prior_depths_grad_threshold to {prior_depths_grad_threshold}.") + # lpips_model = lpips.LPIPS(net="alex", spatial=True).to(device) + # breakpoint() + print("fg Gaussian num:", renderer.d_model.N) + print("bg Gaussian num:", renderer.s_model.N) + print("generate out_dicts_t.pkl:") + for i, (w2c, t) in enumerate(zip(tqdm(w2cs), ts)): + assert i == t.item() + # if i<98: + # continue # test + with torch.inference_mode(): + # out_dict=renderer.model.render(int(t.item()), w2c[None], K[None], img_wh) # my code + gs5 = [] + if renderer.bg_flag: + gs5.append(renderer.s_model()) + if renderer.fg_flag: + d_list_t=renderer.d_model(t) + gs5.append(d_list_t) + n_fg = renderer.d_model(t)[0].shape[0] + render_dict, out_dict = render_gsplat( + gs5, + H, + W, + K=K, + T_cw=w2c, + bg_color=[1.0, 1.0, 1.0], + n_fg=n_fg, + ) + n_all=len(gs5) + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # density filter begin + mu, frame, scale, o, sph=d_list_t + # S = torch.zeros_like(frame) + # S[:, 0, 0] = scale[:, 0] + # S[:, 1, 1] = scale[:, 1] + # S[:, 2, 2] = scale[:, 2] + # actual_covariance = frame @ (S**2) @ frame.permute(0, 2, 1) + inv_actual_covariance = frame @ torch.diag_embed(1.0 / (scale ** 2)) @ frame.transpose(1, 2) + + # test = torch.distributions.MultivariateNormal(mu, actual_covariance).log_prob(mu) + density=gaussian_kde_density_knn(mu,inv_actual_covariance, k=100) # TODO: check scale factor + # breakpoint() + large_density_Gaussian_strict=density>thr_density_strict # TODO: threshold # 5 + large_density_Gaussian_loose=density>thr_density_loose # TODO: threshold # 1 + # density filter end + + out_dict['w2c']=w2c.detach() + out_dict['K']=K + # gt_mask = renderer.s2d.dep_mask[i].to(device) # (H, W) # test # not the true mask! + gt_mask = gt_masks[i] + # breakpoint() + # out_dict['gt_mask']=gt_mask + projected = project_points(out_dict['means'], K, w2c) # [N,2] + in_mask_Gaussians=valid_and_visible(projected, gt_mask) # [N] bool + out_dict['in_mask_Gaussians']=in_mask_Gaussians + contribs=out_dict['contribs'].reshape(-1,3)[:,0].clone() + + img = out_dict["img"][0] # [H,W,3] + gt_img = renderer.s2d.rgb[t].to(device) # [H,W,3] + diff_img = F.l1_loss(img, gt_img, reduction="none").mean(dim=-1) # [H,W] + mask_diff_img = diff_img < thr_rgb_diff # [H,W] # TODO: threshold #0.1 + small_img_diff_Gaussians=valid_and_visible(projected, mask_diff_img) # [N] bool + + # ssim map begin + ssim_map, ssim_loss=compute_ssim_map(img.cpu().numpy(),gt_img.cpu().numpy()) # [H,W,3] + mask_ssim_img = torch.tensor(np.array(ssim_map > thr_ssim),device=device).all(dim=-1) # [H,W] # TODO + large_ssim_Gaussians=valid_and_visible(projected, mask_ssim_img) # [N] bool + + # lpips_map, lpips_loss=compute_lpips_map(lpips_model, img.cpu().numpy(),gt_img.cpu().numpy(),device=device) # [H,W,1] + # mask_lpips_img = torch.tensor((lpips_map < 0.5)[:,:,0],device=device) # [H,W] # TODO + # small_lpips_Gaussians=valid_and_visible(projected, mask_lpips_img) # [N] bool + # breakpoint() + # ssim map end + # depth gradient mask + mask_depth_grad_img = prior_depths_grad[i] < prior_depths_grad_threshold # [H,W] + small_depth_grad_Gaussians=valid_and_visible(projected, mask_depth_grad_img) # [N] bool + + # + valid_gaussians_strict = small_depth_grad_Gaussians & large_ssim_Gaussians & small_img_diff_Gaussians & in_mask_Gaussians # & large_density_Gaussian_strict #& small_lpips_Gaussians # & in_mask_Gaussians # + valid_gaussians_loose = large_ssim_Gaussians & small_img_diff_Gaussians # & large_density_Gaussian_loose #& small_lpips_Gaussians # + # valid_gaussians=torch.ones(n_all, dtype=torch.bool, device=device) # TODO: threshold + # contribs_in_mask[~small_img_diff_Gaussians]=1e-8 # Gaussians on incorrect pixel are set to small contribs + contribs_in_mask=contribs.clone() + contribs_in_mask[~in_mask_Gaussians]=1e-8 # out-mask Gaussians are set to small contribs + contribs_strict=contribs.clone() + contribs_loose=contribs.clone() + contribs_strict[~valid_gaussians_strict]=1e-8 + contribs_loose[~valid_gaussians_loose]=1e-8 + + out_dict['contribs_in_mask']=contribs_in_mask + out_dict['contribs_strict']=contribs_strict + out_dict['contribs_loose']=contribs_loose + + # depth uncertainty + depths_info_perG=associate_pts_value(projected, depths_info[i]) # per Gaussian [N] + out_dict['depths_info_perG']=depths_info_perG + # out_dict['prior_depths_grad']=prior_depths_grad + + if True: + import pickle + dict_dir=f"{save_dir}/out_dicts3" + os.makedirs(dict_dir, exist_ok=True) + file=open(f'{dict_dir}/out_dict_{t}.pkl', 'wb') + pickle.dump(out_dict, file) + file.close() + + if True: + def to_rgb(img: np.ndarray) -> np.ndarray: + # [H, W] or [H, W, 1] -> [H, W, 3] + if img.ndim == 2: + return cv2.cvtColor(img, cv2.COLOR_GRAY2BGR) + elif img.ndim == 3 and img.shape[2] == 1: + return cv2.cvtColor(img[:, :, 0], cv2.COLOR_GRAY2BGR) + return img # Already RGB or unsupported format + + img_np = (img.cpu().numpy() * 255.0).astype(np.uint8) + gt_img_np = (gt_img.cpu().numpy() * 255.0).astype(np.uint8) + ssim_map_np = (np.array(ssim_map) * 255.0).clip(0, 255).astype(np.uint8) + # lpips_map_np = to_rgb((np.array(lpips_map) * 255.0).clip(0, 255).astype(np.uint8)) + + img_with_points = draw_projected_points_masked(img_np, projected, valid_gaussians_strict) # valid_gaussians, in_mask_Gaussians + img_with_points_loose = draw_projected_points_masked(img_np, projected, valid_gaussians_loose) + masked_img = overlay_mask_yellow(img_np, gt_mask.cpu().numpy()) + masked_gt_img = overlay_mask_yellow(gt_img_np, gt_mask.cpu().numpy()) + masked_diff_gt_img = overlay_mask_yellow(gt_img_np, mask_diff_img.cpu().numpy(), color= (255,0,0)) + + diff_img_np = ((diff_img*5).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # *10 to make it clear + diff_img_np = to_rgb(diff_img_np) + # depth_info_np= ((100*depths_info[t]).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # TODO: auto normalize for show + # depth_info_np = to_rgb(depth_info_np) + img_with_points_depth=draw_projected_points_colored(img_np, projected, depths_info_perG) + depth_img=apply_depth_colormap( + prior_depths[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = to_rgb(depth_img_np) + prior_depths_grad_img=apply_depth_colormap( + prior_depths_grad[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + prior_depths_grad_img_np = (prior_depths_grad_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + + # hist_img = plot_density_histogram_rgb(density,height=img_np.shape[0]) + + # combined = stack_images_side_by_side(img_with_points, masked_img) + # lpips_map_np, masked_img, masked_gt_img, hist_img , depth_info_np, depth_img_np, + combined = np.hstack([img_with_points, img_with_points_loose, depth_img_np, prior_depths_grad_img_np, ssim_map_np, masked_img, masked_gt_img, masked_diff_gt_img, diff_img_np, img_with_points_depth]) + iio.imwrite(f"{video_dir}/img_{i}.png", combined) + # breakpoint() + # del lpips_model.net + # lpips_model=None + # Delete large objects manually + del renderer + del render_dict + del inv_actual_covariance + # del lpips_model + del prior_depths + del prior_depths_grad + out_dict = {k: v.detach().cpu() if isinstance(v, torch.Tensor) else v for k, v in out_dict.items()} + del out_dict + del img, gt_img, diff_img, mask_diff_img + del projected, depths_info_perG + del contribs_strict, contribs_loose + del ssim_map, ssim_loss#, lpips_map, lpips_loss + + + # Optionally clear lists + gs5.clear() + + # Force garbage collection + import gc + gc.collect() + + # Clear PyTorch CUDA cache + torch.cuda.empty_cache() + # breakpoint() + +def compare_images(img, img0, tolerance=1): + if img.shape != img0.shape: + print(f"Shape mismatch: {img.shape} vs {img0.shape}") + return False + + diff = np.abs(img.astype(np.int32) - img0.astype(np.int32)) + max_diff = diff.max() + + if max_diff <= tolerance: + print(f"Images are close. Max diff: {max_diff}") + return True + else: + print(f"Images differ. Max diff: {max_diff}") + return False + +# def test(): +# # iphone +# # seq_name="block" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five, spin +# seq_name="backpack_" #"backpack_" +# # work_dir=f"/data/dataset/iphone_mosca_trained/paper-windmill/logs/iphone_fit_native_add3_20250226_115418" +# work_dir=f"/data/dataset/iphone_mosca_trained/backpack_/logs/iphone_fit_native_add3" #_20250223_123930" +# save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" +# dataset_name="iphone" + +# file_name_graph=f"v12_parameters8mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) +# main3(save_dir,file_effective,file_name_graph) + +# def test3(): +# # davis +# seq_name="camel" # rhino, camel, car-turn +# work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" +# save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" +# # root_dir=f"/data/dataset/custom_mocsa_initialized" +# dataset_name="davis" + +# file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 +# main3(save_dir,file_effective,file_name_graph) + + +def test2(): + # iphone + # seq_names=['paper-windmill', 'block', 'teddy', 'apple', 'haru-sit']#, 'sriracha-tree', 'mochi-high-five', ] #,'sriracha-tree','block','teddy'] + # seq_name="mochi-high-five" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five + # 'backpack_', + # seq_names=['backpack_', 'apple', 'block', 'handwavy_', 'mochi-high-five_', 'pillow_', 'spin', 'teddy', 'creeper_', 'haru-sit_', 'paper-windmill', 'space-out', 'sriracha-tree_', 'wheel'] + # seq_names=['apple', 'block', 'spin', 'teddy', 'space-out', 'wheel'] + seq_names=['apple'] + + depth_ratio=0.1 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + # auto find the folder with timestamp + # base_dir = f"/data/dataset/iphone_mosca_trained/{seq_name}/logs" + # pattern = os.path.join(base_dir, "iphone_fit_native_add3*") + # matches = glob.glob(pattern) + # if matches: + # work_dir = matches[0] # or choose with more logic + # print("Found:", work_dir) + # else: + # print("No matching folders found.") + # print('='*20) + # continue + + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/logs/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + dataset_name="iphone" + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + # main1(dataset_name,work_dir,save_dir,ws) + torch.cuda.empty_cache() + file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="max_contrib",threshold_min=0.5) #"max_contrib" minmax_distance_contrib + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + + + +def test4(): + # davis # 'car-turn', 'camel', + # seq_names=['car-turn', 'bear', 'breakdance-flare', 'dance-twirl', 'drift-straight', 'helicopter', 'libby', 'mbike-santa', 'rhino', 'schoolgirls'] # + seq_names=['camel'] + # seq_names=['car-turn', 'bear', 'drift-straight', 'helicopter', 'mbike-santa', 'rhino'] # + depth_ratio=0.001 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + dataset_name="davis" + + file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning + # main1(dataset_name,work_dir,save_dir,ws) + file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="minmax_distance_contrib",threshold_min=0.5) #"minmax_distance_contrib") + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + +def backup_code(save_dir): + from datetime import datetime + backup_dir = osp.join(save_dir, "src_backup") + os.makedirs(backup_dir, exist_ok=True) + for path in [ + "run_preprocessing.sh", + "graph_model_preprocessing5.py", + ]: + os.system(f"cp -r {path} {backup_dir}") + # reduce the backup size + # shutil.rmtree(osp.join(backup_dir, "lib_prior", "seg")) + # for root, dirs, files in os.walk(backup_dir): + # for file in files: + # if file.endswith(".pth") or file.endswith(".ckpt"): + # if osp.isfile(osp.join(root, file)): + # os.remove(osp.join(root, file)) + # else: + # shutil.rmtree(osp.join(root, file)) + # backu the commandline args + with open(osp.join(save_dir, f"commandline_args{datetime.now().strftime('%Y-%m-%d-%H%M%S')}.txt"), "w") as f: + f.write(" ".join(sys.argv)) + +def run_from_outside(): + parser = argparse.ArgumentParser() + parser.add_argument("--seq_name", type=str, required=True) + parser.add_argument("--func_name", type=str, required=True) + parser.add_argument("--dataset_name", type=str, required=True) + parser.add_argument("--depth_ratio", type=float, help="default=0.1 (iphone) or 0.001 (backpack or davis)", required=True) # default=0.1, + parser.add_argument("--threshold_min_set", type=float, help="default=0.5 or 0.1 (paper-windmill)", required=True) # default=0.5, + parser.add_argument("--version_key_edges", type=str, help="version_key_edges: max_contrib or minmax_distance_contrib", required=True) # default="max_contrib" + parser.add_argument("--extra_save_str", type=str, help="extra_save_str", required=False, default="") # _ab_uknn ## ablation study + parser.add_argument("--ratio_key", type=float, help="ratio of key nodes to determine the selected number of nodes in the first stage and threshold in the 2nd stage", required=False, default=-1) + parser.add_argument("--which_logs_folder", type=str, help="which_logs_folder", required=False, default="logs") # default="logs" + parser.add_argument("--log_subfolder", type=str, help="log_subfolder", required=False, default="demo_fit_native_add3") # default="demo_fit_native_add3" + # parser.add_argument("--use_ugraph", action="store_true") + parser.add_argument("--relative_dir_saved_mosca_model", type=str, help="relative_dir_saved_mosca_model. e.g.,", required=False) + args = parser.parse_args() + + seq_name=args.seq_name + dataset_name=args.dataset_name + depth_ratio=args.depth_ratio # 0.001 + threshold_min_set=args.threshold_min_set # 0.5 # paper-windmill 0.1 + version_key_edges=args.version_key_edges + extra_save_str=args.extra_save_str + ratio_key=args.ratio_key + # save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}/' # original + save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}{extra_save_str}/' + logs_folder=args.which_logs_folder + relative_dir_saved_mosca_model=args.relative_dir_saved_mosca_model + log_subfolder=args.log_subfolder + if dataset_name=="iphone": + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/{logs_folder}/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + voxel_size=0.2 + n_min_effective_frames=5 + # version_key_edges="max_contrib" + elif dataset_name=="davis": + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # version_key_edges="minmax_distance_contrib" + elif dataset_name=="nvidia": + work_dir=f"/data/dataset/nvidia_mosca_trained/{seq_name}/{logs_folder}/nvidia_fit_native_add3" + save_dir=f"/data/dataset/nvidia_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/nvidia_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=3 + elif dataset_name=="diffusion4d": + work_dir=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/diffusion4d_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # elif dataset_name=="objaverse": # old + # group_name="000-001" + # # /data/dataset/objaverse_mosca_trained/000-001/3ae0b4e6a75a41d38a2adf8db048a703/000/mono/logs/demo_fit_native_add3 + # work_dir=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/{logs_folder}/demo_fit_native_add3" + # save_dir=f"/data/dataset/objaverse_mosca_graph_model/{group_name}/{seq_name}/000/mono/"+save_sub_folder + # ws=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/" + # voxel_size=0.05 + # n_min_effective_frames=5 + elif dataset_name=="objaverse": + work_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/{logs_folder}/{log_subfolder}" # TODO demo_fit_native_add3_0 + save_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_graph_model/{seq_name}/mono/"+save_sub_folder + ws=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/" + voxel_size=0.1 # TODO + n_min_effective_frames=20 # TODO + elif dataset_name=="davis_mask": + work_dir=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model_masked_4x/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + else: + raise ValueError("dataset_name not supported") + + backup_code(save_dir) + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + + if args.func_name=="main1": + main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model) + elif args.func_name=="main23": + if args.depth_ratio is None or args.threshold_min_set is None or args.version_key_edges is None: + parser.error("--depth_ratio, --threshold_min_set, and --version_key_edges are required when func_name is 'main23'") + file_effective=main2(save_dir,voxel_size=voxel_size,n_min_effective_frames=n_min_effective_frames,ratio_key=ratio_key) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges=version_key_edges,threshold_min_set=threshold_min_set) #"max_contrib" minmax_distance_contrib + + + +if __name__ == "__main__": + # test2() + # test4() + # DON'T FORGET TO CHANGE! + run_from_outside() \ No newline at end of file diff --git a/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh new file mode 100644 index 0000000000000000000000000000000000000000..fade29ad45ed691be51bd485f7dd3274bb62fad1 --- /dev/null +++ b/000-000-00616f328a8b4b5a8e689f61e70758b6-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh @@ -0,0 +1,159 @@ +#!/bin/bash + +gpu_id=3 +dataset_name=iphone + +# 'backpack_' 'haru-sit_' 'paper-windmill' +seq_names=('paper-windmill' 'block' 'teddy' 'wheel' 'apple' 'spin' 'space-out') +# seq_names=('teddy' 'wheel') +# seq_names=('spin' 'space-out') +# seq_names=('paper-windmill') +# seq_names=('block') +# seq_names=('haru-sit_' 'handwavy_' 'mochi-high-five_' 'sriracha-tree_' 'creeper_' 'pillow_') +# seq_names=('haru-sit_') # 'backpack_' + +depth_ratio=0.1 +# threshold_min_set=0.1 # 0.1 paper-windmill # 0.5 others +threshold_min_set_dict=( + ["paper-windmill"]=0.1 + ["teddy"]=0.5 + ["wheel"]=0.5 + ["apple"]=0.5 + ["spin"]=0.5 + ["space-out"]=0.5 + ["haru-sit_"]=0.5 + ["handwavy_"]=0.5 + ["mochi-high-five_"]=0.5 + ["sriracha-tree_"]=0.5 + ["creeper_"]=0.5 + ["pillow_"]=0.5 +) +version_key_edges=max_contrib # max_contrib minmax_distance_contrib max_native +ratio_key=0.02 # default -1 +extra_save_str=_ab_key_ratio_$ratio_key #2 + +# Loop and print each name +for seq_name in "${seq_names[@]}" +do + echo "${seq_name}" +done + +# Loop over each sequence +for seq_name in "${seq_names[@]}" +do + echo "Starting training for ${seq_name}" + + threshold_min_set=${threshold_min_set_dict[$seq_name]} + + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main1 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str # comment if no extra_save_str + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main23 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str \ + --ratio_key $ratio_key # ablation study + + echo "Finished training for ${seq_name}" + echo "*****************************************" + echo "*****************************************" +done + + + + + + +# gpu_id=2 +# dataset_name=davis +# seq_names=('camel' 'car-turn' 'drift-straight' 'helicopter' 'mbike-santa' 'koala' 'parkour' 'sheep' 'soccerball' 'schoolgirls') # 'bear' 'rhino' +# # seq_names=('camel' 'train' 'car-roundabout' 'car-turn' 'bike-packing' 'drift-straight' 'helicopter' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'mbike-santa' 'parkour' 'pigs' 'rhino' 'schoolgirls' 'sheep' 'soccerball' 'swing' 'bear') +# # seq_names=('train' 'car-roundabout' 'bike-packing' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'parkour' 'pigs' 'schoolgirls' 'sheep' 'soccerball' 'swing') + +# depth_ratio=0.001 # 0.1 +# threshold_min_set=0.1 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done + + + +# gpu_id=1 +# dataset_name=nvidia + +# # seq_names=('Balloon1' 'Balloon2' 'Jumping' 'Playground' 'Skating' 'Truck' 'Umbrella') # +# # seq_names=('Balloon1' 'Jumping' 'Truck' ) # 'Umbrella' +# seq_names=('Balloon2' 'Playground' 'Skating') + +# depth_ratio=0.1 +# threshold_min_set=0.01 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done \ No newline at end of file diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/004.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/004.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/004.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/011.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/011.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/011.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/015.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/015.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/015.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/018.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/018.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/018.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/019.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/019.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/019.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/020.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/023.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/024.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/028.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/028.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/028.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/034.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/034.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/034.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/036.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/036.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/036.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/039.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/039.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/039.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/040.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/043.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/043.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/043.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/050.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/053.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/053.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/053.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/059.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/059.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/059.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/063.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/063.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/063.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/066.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/066.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/066.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/072.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/072.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/072.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/076.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/076.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/076.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/084.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/084.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/084.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/091.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/091.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/091.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/092.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/099.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/099.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/099.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/100.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/102.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/102.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/102.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/104.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/104.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/104.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/111.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/111.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/111.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/114.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/114.png new file mode 100644 index 0000000000000000000000000000000000000000..a05d273e18e29e28506ed3041e7fd8a815ea2ba7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mask_square/114.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19db628a2d921d030da5ee84f447f293688354c9bae8d566c86393a60d1fccbc +size 1074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..8f138ecefd104d19ab5b636fc4ff0dcb7b1b4cf1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da7d38798561ca060d7cf2f5e852d72dca8add75dbc1a0ca8ec6004e4915e4ec +size 199427 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..3a9a74d284f54157a164b18ab408dcb72cbf64a7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54db1cbc68254e66cc185a507d9715864cc32a444d3887e5d1e6f2aee0ec27c4 +size 64515 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_002.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_002.png new file mode 100644 index 0000000000000000000000000000000000000000..f8107c049b2fd8e8c6b1c81a5ffa0b5f848f3fae --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_002.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:450d39482b1bb67147e4a5640618399348099e9ccf9ace563a00cfb7a74ddb8e +size 65312 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_003.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_003.png new file mode 100644 index 0000000000000000000000000000000000000000..8566a67ae4d03b70a7638bd2825c3696f1cf7a98 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_003.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c692e0f3db9c3300b1a8a0bda17f789cd6cc2001af31c70f3f71bcb1e00a2b93 +size 68000 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_004.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_004.png new file mode 100644 index 0000000000000000000000000000000000000000..128e51826921649ebb69e737bb506f44ca2cd123 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_004.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:23c0b997b2cdbffcf628aa48e646468f68ad9680b7bc2168aa07bcc5ba0ac4ae +size 69737 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..6d739201ccbc91c62df0311009fccf3393e80dd9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c2ac41e9a9b0c2b811f42942936c3418b9d3dabc26af6423ff44beb5d5b5bd3 +size 70362 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_008.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_008.png new file mode 100644 index 0000000000000000000000000000000000000000..ad73168c590f94b9f00c8e01874b6080af6b2cff --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_008.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0427c46a130aa1ce12880dba1716a4b873829368f54a82b2563bf2cb34a9b024 +size 70935 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_009.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_009.png new file mode 100644 index 0000000000000000000000000000000000000000..7d76551e1e7ce47d069f7e51d1ffe3e3cc639571 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_009.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c35b034399c575edfdf5e278d273fa3442402e95b1d80b3095debcd4d215bb5e +size 70975 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_010.png new file mode 100644 index 0000000000000000000000000000000000000000..82e03868ccb4b4740a19b31736e30708b3bebb73 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:191d6c73a8b2abd16fa2524d599da7ffbd66e9978759336c3bc3dd2ea6b3adda +size 71086 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_011.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_011.png new file mode 100644 index 0000000000000000000000000000000000000000..0853f6b7b7430eeb29449ba176a7a024eaad08e4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_011.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7488a91ac29d36444fe2d6cfded586e74d6caae16fd5a4d4a034f41bbe70c271 +size 70578 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_013.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_013.png new file mode 100644 index 0000000000000000000000000000000000000000..6e1dceb7fed41268e04cbca4da369548d21dc148 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_013.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58813a743449c488b49f662748d0425fd1f7c600b520380104d506766460952d +size 70467 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_014.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_014.png new file mode 100644 index 0000000000000000000000000000000000000000..935e1dc8cd3826ef1393a0d7850c0fe43a6f0694 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_014.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04d43053038278ac84ccead9062cb1c9f6af77c7917b7d2abaee29ad79b89db8 +size 70160 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_015.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_015.png new file mode 100644 index 0000000000000000000000000000000000000000..de5a0f890078645550091d9cc94ccb9d66746efc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_015.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:774ac258f9c15eeb6d913ac12796633520c077801c567f60ffea82ee189e0071 +size 69530 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_016.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_016.png new file mode 100644 index 0000000000000000000000000000000000000000..f01735357b193c66a049509811c6eeb8df560a19 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_016.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95a1926d575b532d23a35064aee1e42bf7352842db7e670b7612011e3d0d1cee +size 69274 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_017.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_017.png new file mode 100644 index 0000000000000000000000000000000000000000..6f5e49572394115e2e22bf9993a7b392aeab9c47 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_017.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:23a8891e1c38fdb75042783101e4662d3bd9df48addcc2e0770377d79cd6f3dc +size 69724 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_018.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_018.png new file mode 100644 index 0000000000000000000000000000000000000000..87731ba29105ba522f5fca21755578914823425a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_018.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1bae9533a7f71a9f4d58080b12dd456262a2697ee183dfe869f08c327735911c +size 69642 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_021.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_021.png new file mode 100644 index 0000000000000000000000000000000000000000..e1c284961816b35ed6aa60430ea9ad9b71965512 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_021.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4742311a6fc163bc0e65771247b4c9820eb3b27a5b5f2d7fb1e37b195f62d5c4 +size 70271 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_022.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_022.png new file mode 100644 index 0000000000000000000000000000000000000000..96a547dc79f028ea0ee65aaf5b3c03ca57e0ed5f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_022.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9338aa0da7ff5aabc2ef251669069d8180671f276e8f4037167474744eba9a7d +size 71062 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_023.png new file mode 100644 index 0000000000000000000000000000000000000000..0f7e7756cf930a2038388f0b1760bf7aa80a0450 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70dc535ee3d599637ef65bb251848d088fdb549f53ff1c629df6453e2104569b +size 72417 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_025.png new file mode 100644 index 0000000000000000000000000000000000000000..8ae98042940a040153230b61e1ccf66981097a07 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5363e6704617f9033fe411ccf303f90d5dbbeea4db5adba2661f7749cdb5db1 +size 74791 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_026.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_026.png new file mode 100644 index 0000000000000000000000000000000000000000..33380abd039a8e99ce3a666628c8bb97035f9488 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_026.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df97a09f1e6cdb5987d801edeefa969caa73b7ba0b8368303dc2d381d7206bd2 +size 75513 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_027.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_027.png new file mode 100644 index 0000000000000000000000000000000000000000..4334ae6d4eaed9fcf494dc55d67db0668d91350a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_027.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85616564997c6ee080145a4d6b347d5491403991ecd12de26c8c4245bce7fab7 +size 76654 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_028.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_028.png new file mode 100644 index 0000000000000000000000000000000000000000..e8e597b4a06ea0487f2f0a5061901f0e373dda2b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_028.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:761e631d58530fdb6a0f4034bfe8d8ea0153477886cf9376fc2a2f314bb656b8 +size 77644 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_029.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_029.png new file mode 100644 index 0000000000000000000000000000000000000000..4bb04934ab27a7e0426e93b649976c819415f6eb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_029.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ecfb091b9255ccaa301a68356b0e978d1e378da7fea49a929762ec5a2f2104c4 +size 77750 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_031.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_031.png new file mode 100644 index 0000000000000000000000000000000000000000..72c7a44f05c343d7f0a3d101e71136ec6f920f46 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_031.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a69216996dc5cc3660166df3b37b70f3994acc95cedac6a239dcd78dc71e1ffb +size 77971 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_034.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_034.png new file mode 100644 index 0000000000000000000000000000000000000000..53cd4b7ad8642cbe2d22d7858c05e0a003f9fd6f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_034.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:caff0c70113cdca848945956b18bb66d5236360f328c7abbf2649355b08efbc9 +size 74331 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..550ef28347807014206859f83dfe6435a44adef3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f536c8c9db71fa3bb710126a967a4fdb1f83ca09289a791ab26e4bfdff4057ea +size 73236 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_037.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_037.png new file mode 100644 index 0000000000000000000000000000000000000000..fcda9d661fa27e194f1f69bd593d3667c2e7883c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_037.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5026c67d33a147943f62f93a28435af2a2d30116d7bb5283520eb6cab357bd1c +size 71204 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_039.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_039.png new file mode 100644 index 0000000000000000000000000000000000000000..105f1e98ebd29ae11d51da2e509e1dd3feeebb1a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_039.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f6fb00a45288a8e6223655291237c0807ce0003b1921d01593235c617fbac6c +size 68637 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_041.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_041.png new file mode 100644 index 0000000000000000000000000000000000000000..da3fac8e03c06d46fa975519f787699e52258e31 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_041.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:30ed10acda937b03290708fa0412ecf8efd08f6802ee975566241c7bc009ebb4 +size 67904 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_042.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_042.png new file mode 100644 index 0000000000000000000000000000000000000000..0d92d71e9e58b5479ed81c2f7ce77ffaa46bbb99 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_042.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ac8e800ac9c16a4d468eacb453cdb0df82fb675d41c03383e7f3ad37f091acd +size 68616 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_043.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_043.png new file mode 100644 index 0000000000000000000000000000000000000000..373c100dcfb4090a95feb905c734512c975703e1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_043.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca12ee8134e56655d546589235793a48d2753f5c8d1fe2fc0541fe3ed6c5822f +size 69370 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_044.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_044.png new file mode 100644 index 0000000000000000000000000000000000000000..8bbc226e6a293d9e1eb341f8303f2e2b477f4e55 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_044.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f014e514d2f5e8817a8bca390826e85825dfa5da493ef90322f66ead0756be86 +size 70609 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_045.png new file mode 100644 index 0000000000000000000000000000000000000000..63887c9e67073820f2ec66e09a4a3448acc8fe5d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e2df7922b332f48908aca0f4d69de8be42cf9289d8e49d96954eddd7f8292b2 +size 71485 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_046.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_046.png new file mode 100644 index 0000000000000000000000000000000000000000..49056c5b009fec404379f64f564040cae57f54dc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_046.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fdbba61de96a6328d0b5991f42149f206320a2cf9bb806e168b40c865b9d995 +size 73020 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_048.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_048.png new file mode 100644 index 0000000000000000000000000000000000000000..8b9caef9d9aa7dd1446c7a17eae910c17366f7d7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_048.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:23968fbf0be9be83dfa4620bc59a1d7b60476898e28abe43f787c0be3990e988 +size 75405 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_049.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_049.png new file mode 100644 index 0000000000000000000000000000000000000000..06dab29390fd93581b008ea003d5dc2086821885 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_049.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:184296575bf1f262a6b73d422daede30e83c8a2d5767741d59c5fb5061f2caeb +size 75356 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..f1fb3140cd1d3e2963c047c2b985fdfd122fbeae --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:188fdef60b4dce3684edafbd53f9e43fc442fadb000c2ec8f4a1b28b57630c11 +size 74915 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_051.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_051.png new file mode 100644 index 0000000000000000000000000000000000000000..ad85825375024f9655191ea4c16a031128b6117a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_051.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:007a38efa5dc12966df7c982679ff89267226435de1bd3cd3380817a51359730 +size 74538 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_052.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_052.png new file mode 100644 index 0000000000000000000000000000000000000000..099e4126104e535c47f136e95846455f5fd7bb7f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_052.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a426efc21267a89778a20c35d91d3e496df5223e3e801fcb4f588b62faae6c1 +size 73967 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_058.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_058.png new file mode 100644 index 0000000000000000000000000000000000000000..cdc1344cfe39bf9fc287a86af362114a4c82b456 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_058.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee5f508499b5df0753177e0c34b9934772dfb0e7dad7b2bfc2ef74921f244c91 +size 68716 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..158b2a8ecf164b73b3f5487fe9f5d4d556dfb8b8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c31cbb93cdc037dec413676ed1a76ab27188334c4e43ebaaf7e41433b08eeb4c +size 68195 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_061.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_061.png new file mode 100644 index 0000000000000000000000000000000000000000..5380a614bb298366933186917b49d924c255d13a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_061.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85ce97969c0657d4f7d7d1f56a3e53c59b76bbae75d83896b6a7cebe3a9fa22e +size 67990 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_063.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_063.png new file mode 100644 index 0000000000000000000000000000000000000000..fc51b22a815c984f24b0139da6620495b8d819c2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_063.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9818b4726574a02b9355c98dd93245c1303652ea6a8bf3b79a0885f2b38ce5bc +size 68252 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_064.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_064.png new file mode 100644 index 0000000000000000000000000000000000000000..9ba00e0de356625876f0682706af014df30483b4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_064.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0cbc16b223c1ff912e1ea0f3b8350ca8d3683f7796656db0e360a528a1e3924d +size 68143 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..ac9771aa9e7a73e420feba58a4e105042e22a470 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:640d4510989eeedb19432631bea9a0c869d21012e68b43dbb5baf1c1192f9f8e +size 68924 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_067.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_067.png new file mode 100644 index 0000000000000000000000000000000000000000..6981b227e61340526e0136d9dec28a85a283ec26 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_067.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb540d445e245c02aa00be06bb0f65be8b170a684b9ec3250a071917a2aaa023 +size 70106 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_069.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_069.png new file mode 100644 index 0000000000000000000000000000000000000000..f753f03f619b2cc31ec98ed5cf09b9814bb14c29 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_069.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8631ff100843e1303194d71af2de117d82c21eea235642ce8fd2430c1c2820e +size 72240 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_072.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_072.png new file mode 100644 index 0000000000000000000000000000000000000000..e248769c783472c78982cde96c2cd68f22f4a6f2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_072.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac7b51aabd07b4ca346fa83fffb2b217c74955ca3c4bf63219857dcfa71e5132 +size 74863 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_073.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_073.png new file mode 100644 index 0000000000000000000000000000000000000000..d07d47e26a9e7d4191686fc2e7a0603d418e77ab --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_073.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58c3f1b9fba0d7eec09e97ff72ad8853e9f2c7a90d66c8c9085008ee41e985be +size 74840 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_074.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_074.png new file mode 100644 index 0000000000000000000000000000000000000000..f49c8bfcd5d894fe4a529b2ca7b6d2152a9dfa76 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_074.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6c97240c0083a9328f1ee9f559fa9259b81c9996b86d13a2c1c9791da93ff37 +size 74396 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_076.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_076.png new file mode 100644 index 0000000000000000000000000000000000000000..ab5867ee73fa24866239d8b6847d5ab5f097ed57 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_076.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb6c98ddf3127ea671a929eed3767d3710e4491d2bee10938e9faae55056dd6b +size 73743 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_078.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_078.png new file mode 100644 index 0000000000000000000000000000000000000000..13e1137b16ec8f4a5fe2cf5ed6a0b4033f47d0c2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_078.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:231ae186b38cb2a0c8c07ef2eaefd3b5b68cb7e23bdada5d7db9b9c9d48274d5 +size 71988 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_079.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_079.png new file mode 100644 index 0000000000000000000000000000000000000000..05d42df22e452575e6603b7f402f97b135c46e3d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_079.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dae962004b70d053a47e44802d757392766ab587b20bd89d8a968c3629c9ff2e +size 71823 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_083.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_083.png new file mode 100644 index 0000000000000000000000000000000000000000..25d9e41eacb4bc185317a15591aabf34c20e44d7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_083.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c6dc54686ca8707a9968b7d5230d2f6aff3c93aaa5f1bcc39445e64cda77b81 +size 72183 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_084.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_084.png new file mode 100644 index 0000000000000000000000000000000000000000..6a2517e35253b31de3ea075d34ddde498e4f695c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_084.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4abef6847d7faaf89afa8a13dd7620aacb652f2cec55fc746ee15ac553e20d9 +size 73365 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_086.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_086.png new file mode 100644 index 0000000000000000000000000000000000000000..c9b3eeff684cd193f386d70b7bb399c446d502ee --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_086.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2aae3354ca43d4eadfb574747b5eced7d04f1bebeef602c6a8899fdcb45cbea +size 76389 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_087.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_087.png new file mode 100644 index 0000000000000000000000000000000000000000..5109dbd9c4d3e040d856d5a9cb6ff9fafb73ad43 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_087.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33bc8c98c44d58902503aec8866e634c3fab2753101a3c79c6e6c13f92521ffb +size 77229 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_088.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_088.png new file mode 100644 index 0000000000000000000000000000000000000000..80122f454ee5ea60af2dccd624087f10b1b60daf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_088.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b52b0d07e3c67e72b90fecfc8ee5b23a3e596e32c9201234dccd041738e3ae9e +size 77312 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_089.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_089.png new file mode 100644 index 0000000000000000000000000000000000000000..ecf7d57b724e2c8805ac6f9245ed415cadf482c3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_089.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85cff77b83bd859d1481d79000fd292dbeb16099f1ee7a6b0d5d58e33350f698 +size 76868 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_093.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_093.png new file mode 100644 index 0000000000000000000000000000000000000000..e6e6e297a2c53c469a5dcff2ff0c4da2072f6a43 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_093.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f73e995bc3f33d81a250935b04b6a6a525e2ec2712e2c61ae8f0db4a237bffdf +size 73604 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_094.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_094.png new file mode 100644 index 0000000000000000000000000000000000000000..6e8f951835b167b3ff54c2b38f701e6ebf0409d8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_094.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:305a9ffd9d6020e8c50c52aa3344ab5a606c0dc25005ae2a818ea41a7f703c2c +size 72468 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..9007a559883506503d811d7b5a1a13b57d0176bc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29ec5026187114608a22b8ea4738dd6a0d5b1308c579b4a9bf1b61346a71c9b2 +size 71617 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_096.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_096.png new file mode 100644 index 0000000000000000000000000000000000000000..1354719b20a68cd506750f2341d6b3d4afb7fc8b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_096.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:281f612c0c053a1cb8443decc9a3b34727ab44374fa8bd4d7d7655d71c2494ba +size 70700 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_097.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_097.png new file mode 100644 index 0000000000000000000000000000000000000000..aee738816fc6a068bb452924235a2147f3a88ee9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_097.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36c2877d4f91a816cf6ad0fa7bf9a06451ff29366bf873150bbd78e8d73f4493 +size 69724 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_098.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_098.png new file mode 100644 index 0000000000000000000000000000000000000000..9343742a3a9249582559b772246f53c51124ecb0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_098.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86c4cdb5ff7e674a83589c3d6c7c20b024c3bb10db8f2dfeb2226e6eb7bf6cbc +size 69265 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_099.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_099.png new file mode 100644 index 0000000000000000000000000000000000000000..a380a3d29cfaffe2fa2e409600f7a3b538c87b20 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_099.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6625e9adec8b3715d258ed8cef402ad33617e8b7b2f4bd16686ba8a232f9471 +size 69074 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..9102402757278cf59b0c2468fab2e900749e25b5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f5aebd9d45cabbffdb5006f83cbf6c7647d238e45f0e25416e5c503a9d00107 +size 68824 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_102.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_102.png new file mode 100644 index 0000000000000000000000000000000000000000..e9bd90d0d1cfd796626f8cfe29fef9f2c995b469 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_102.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:976fc35a6a4b136435f3c85167f6c8d476e48ad682194249d66a7cd76bced437 +size 68661 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_104.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_104.png new file mode 100644 index 0000000000000000000000000000000000000000..bb145e1b249d347a86955404a958391fffaad816 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_104.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:171a05deb072c0741900eedd8a07eb55a1774ce54524e321374d4714f2feb644 +size 68795 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..55e835c0c75b486cfb0ba00fe1d1d0d2173c3773 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:436ff2a0ad9449b07f8eea04228ba5ee3521f534495930140f8bce1207b5f413 +size 70052 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_106.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_106.png new file mode 100644 index 0000000000000000000000000000000000000000..d7f113c27472b8ebc18cee97caec75ab6cabdbcf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_106.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7e22e32a654515375025ebe067388f0fc8cb95de63095ee0760cd88399e5eb9 +size 70859 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_107.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_107.png new file mode 100644 index 0000000000000000000000000000000000000000..b3dc7612cd0d039b304299f42fef458315a5f0c9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_107.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:902150d03c1bfff55cc6f8999b9244662a321db475e3d4f468a5f6fb3b0e61b9 +size 72391 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_108.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_108.png new file mode 100644 index 0000000000000000000000000000000000000000..08eb13ec819f069248e7a540168f7a739b84011d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_108.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75c2206c5180bc5360b00369afe567310d127b84306d3b6f35391a487ba602c7 +size 73276 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_110.png new file mode 100644 index 0000000000000000000000000000000000000000..2bec893716003d7281776595446554c041442b3f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:250e61c84191579fa94bab53b65891e60de46f49da0e0ab6146a9cd20319bc24 +size 75466 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_114.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_114.png new file mode 100644 index 0000000000000000000000000000000000000000..c5922480bf56c02e30847dbb42fb6d584f8de806 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_114.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3376be3194d4f5c0afb1ba22897c0f5c8db2181b00a3971c2e8c30077137328c +size 68819 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_116.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_116.png new file mode 100644 index 0000000000000000000000000000000000000000..7152422b67b8f47288964372e89ee610ad05c370 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_116.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aea39dded7ce8b0d632f277a6ccb256a795458662a6c2dcba0ec6cb2be553266 +size 65806 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_117.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_117.png new file mode 100644 index 0000000000000000000000000000000000000000..91ae190cf46e546246790facf993719ed5a8689c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_117.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67e8486c2c5627d2203b4621ab2b017c72c7e5c93b24008ea080174fb14381e4 +size 64555 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_119.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_119.png new file mode 100644 index 0000000000000000000000000000000000000000..2e4515953fbe2ad47ce3b260846fb106775f4e24 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca/frame_119.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4db43d46fc586b6e4f1cbc4e9efe3f93ccc429564ecb8f0bc50197ed2aa1668f +size 63218 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..52388248b86cb6180529ba41bc9f929aeef4b2a3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:34a4f53b4d1dd942ec76122f526fd805af0cfd0bdf8dd936703df0ba25c7a5d6 +size 183141 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_001.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_001.png new file mode 100644 index 0000000000000000000000000000000000000000..5feba80e483184f6be3aa160bab56569460feea1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_001.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96fa271794a1d9e0a664214b35fefa763e857f2ce85c81e637d68f861b0ed8c1 +size 55644 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_003.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_003.png new file mode 100644 index 0000000000000000000000000000000000000000..808a691fd46a4d3f231ad4da86fbfb275f362440 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_003.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b308822337b2b360a95203ccaa8dc060af1df13ccea280a67a5aec0e70299ad6 +size 57939 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_004.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_004.png new file mode 100644 index 0000000000000000000000000000000000000000..93dce3645d0aa22f8613ebe752feb06d5dd43ebc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_004.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:587ea70b7a11d8a9b4bd488ca56c0f63463e0039c6b9c95238dbc80d49b20715 +size 59027 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_007.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_007.png new file mode 100644 index 0000000000000000000000000000000000000000..f01c9d56f74b8850cfc001740e82aff793ffba1b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_007.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed7ad71f9b47def6bd9e6525af85cefb8ff648a5dd4848cb769c2dc1fc8c7d9b +size 59390 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_008.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_008.png new file mode 100644 index 0000000000000000000000000000000000000000..26e959c2d46950ea516d8e6f5185e3718e338eb6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_008.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:809889172441cb085a85751610419353707a5640258453e3de7e765d66b15161 +size 59482 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_009.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_009.png new file mode 100644 index 0000000000000000000000000000000000000000..4e62f8a84073604177f1a47ce6e14632ce787ad1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_009.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3880123fbd676c093f096f552b433448af6ddb5af202c12e65f262fa8357e411 +size 59072 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_011.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_011.png new file mode 100644 index 0000000000000000000000000000000000000000..6161e9570dd4a47d32b2219bd0d9501a6d6ed06c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_011.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b750b9c479be6492a59b7a37242b94d3bd79fcfd322a963a5527db60aa278af +size 58511 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_013.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_013.png new file mode 100644 index 0000000000000000000000000000000000000000..d76610d83e79433b3e26dcd0edf8460909114fe8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_013.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:986ffbad086f77579d29688a93bafff50944c1eb7d123dff5fe85f6f8cbbb387 +size 57276 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_014.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_014.png new file mode 100644 index 0000000000000000000000000000000000000000..539b43f8c7f7a494925cf2879804ff8ca640326a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_014.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99daa6bfb668bedf802abdfc52f8feccbfd4b0c3e14e7d71775bd9dad90f8155 +size 56739 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_018.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_018.png new file mode 100644 index 0000000000000000000000000000000000000000..a030b793ee621da3d04f754539e7a74fc0db5b21 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_018.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8fe5be1cb00ca14c1b76e472a7974dc91c989c1d132308e2f5633a8934bf3d13 +size 55488 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_019.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_019.png new file mode 100644 index 0000000000000000000000000000000000000000..5701aa94255b4b4a0670651bd4b30949843614ca --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_019.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:753bfd49ae790ab8a53579daa2f8e74090d82cd84db8be7feee1a605d818331c +size 55726 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..7a5abf1f276d3829b9675c005bd0ee13bfd1b0d5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:785d203f4324a9dea31784dbdf209515eb9d4d6a64996ad3e0059e30662ab4fd +size 55760 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_021.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_021.png new file mode 100644 index 0000000000000000000000000000000000000000..6007683469d04e86083f734acbc73efa2c605e9f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_021.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:871d9e8a1006e3c87689a44be3e7b6461dae93df58ddee58a0b97a56188eb2e4 +size 55975 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_022.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_022.png new file mode 100644 index 0000000000000000000000000000000000000000..db01a67d8b5de7df622cf14d29c051486568165f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_022.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c67b6d923355dc945ba9092d516a63f38a371bea8918c299dfbf944fe03c5af +size 56394 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_023.png new file mode 100644 index 0000000000000000000000000000000000000000..3a834095dc308e43873d5a607473f0bdfd023be9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ee105fcfe6254b0f1a8ea6327434528f796ff69fd60f095e81a77ff74f9b0b9 +size 57113 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_024.png new file mode 100644 index 0000000000000000000000000000000000000000..96c0b189f8095574a0823391ea4a3c452eb32533 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b579dcfe8c2011bfb4605671d8b4036e9046bbd348f4f347b45c424836e5bf08 +size 58007 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_027.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_027.png new file mode 100644 index 0000000000000000000000000000000000000000..666c14f17b36f18b056c4b0b9344e0999cea8219 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_027.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d2fcb39d4c98b60454a67702db8f53be54654d99523b5a5a153d8e23e10450e +size 61407 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_028.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_028.png new file mode 100644 index 0000000000000000000000000000000000000000..9414a3cf76e98891aefd5b66de754658e80b5182 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_028.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11cf593397a9cea7b3cbe36ac1e0f849a0b8f6dbf597fc06ef69c1e7f74af37b +size 62880 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_029.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_029.png new file mode 100644 index 0000000000000000000000000000000000000000..00bf8134ec502545a0d9bebcfa059795c54fae19 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_029.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27c6f7db0bae1cb4bc289b045ba657d859995b128c480dfd099bdfd8916952e2 +size 63097 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_033.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_033.png new file mode 100644 index 0000000000000000000000000000000000000000..476214ef19d220e3ab83f906232851cdc6d6a60e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_033.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:494742b2921c2539e92ecec611bda29b8f9a728b52948465fa9378f9142b10af +size 61376 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_034.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_034.png new file mode 100644 index 0000000000000000000000000000000000000000..132f26eec3ec11c0b2e06fab0e15832610a81b09 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_034.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9bad89ab3415d979f9411819c14a38ffa952acd68c4618ae0fbfd95faf648b7 +size 60200 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..38ac4b0a49aae511ed430f3238ea923c05eec18a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dbf00e887db788e6a2c3bbf9e60d3370aa584cbdd625c9e36760fceeea1a4367 +size 59674 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_036.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_036.png new file mode 100644 index 0000000000000000000000000000000000000000..fa5d9dfecda6e7f0334a6e338cbdf2882f2736b4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_036.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f21b5b7b306acfde530539f919a6e072bb23b26a5d0c08a9ac0c6973c2221af +size 59312 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_037.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_037.png new file mode 100644 index 0000000000000000000000000000000000000000..3a75cdedb92deb0a876226f556a3d86d2ed2e0fe --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_037.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f0e73cbbb5c310f948ff854997ba024e50ede09020ebe2cc40070504a0d8c9a +size 58379 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_038.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_038.png new file mode 100644 index 0000000000000000000000000000000000000000..08de55ed3abe3a76e8676e379a50ee31d1cd29e7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_038.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e212cfa900cce601dcc9555444ef4509cfb7cba650410165c89f6c498a0a1ed +size 57569 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_039.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_039.png new file mode 100644 index 0000000000000000000000000000000000000000..a9c8a336ae83009653a199010e9b66b0ae2a9a72 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_039.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4d45abb99423081a6f13888c4db84100275d584826ba748c7e2da38a4f781b5 +size 56599 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..e2dfdbce3f3516947e2883994e59a2d653d441f8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da17afe55b89ef99419d4b77925909f0639b52e18b43095a60c0f97937b2fd49 +size 56034 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_041.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_041.png new file mode 100644 index 0000000000000000000000000000000000000000..209ba3ec7bdcf15eef247ba2fd82f9b84e7dfa82 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_041.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1078fecff4edfe2a1b27005d13a52fd39e9d2d8718062f5f02ba20d0670e1319 +size 56345 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_042.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_042.png new file mode 100644 index 0000000000000000000000000000000000000000..794965e381d1ad8d6398b9ed43018ee04948840f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_042.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2a84e840c277c36f5f5cdfd5cb376c76c1c2b850e8075fa39b94eee7b9e2f0b +size 57216 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_043.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_043.png new file mode 100644 index 0000000000000000000000000000000000000000..bce81930e1abd435caf2cb91b8880862a8002dd9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_043.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5b7cef768da349494e4dabbbcf46cf41ef9a179f7e9b3d8a1580afe11592f79 +size 57704 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_047.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_047.png new file mode 100644 index 0000000000000000000000000000000000000000..05b8815449a772b1270b077bd061081669b9b4f0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_047.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a286166d9761827141ee06a114685715d88236bfa3f89c0a718afa596b11f56 +size 61920 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_048.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_048.png new file mode 100644 index 0000000000000000000000000000000000000000..5c267e50f67946ab4dcfce06911bd4da305e327f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_048.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f644c3cc80a6f3025c083341d12781356f8d7d70bc972550befa7a931510cfc5 +size 62912 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_049.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_049.png new file mode 100644 index 0000000000000000000000000000000000000000..7c7cd70552dfdc0aa51ccb7bf1ef3b8ac8adc9fd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_049.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff04e7195a9c38e28fc0dbf97221fefed26208e9864dea907165bd89fb656f70 +size 63068 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..f89fca62b4a2cb544f6969de18419b13bd2fe587 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bcbc548cb72c9ae6cd8be6d6ebc051f46640580748f57864662c08cb55c755e1 +size 62907 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_051.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_051.png new file mode 100644 index 0000000000000000000000000000000000000000..4875a8fb91627f8e633dbb5cbf030ece00679a67 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_051.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e5c4dc3262b46832a50a1a741f39ec76fcc7b305c89d75ac89e497810805ea0 +size 63078 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_052.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_052.png new file mode 100644 index 0000000000000000000000000000000000000000..687dd90be7aaf4a3f2bb61e408a5756abbd7cc63 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_052.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c1a95acb68b03c31dc20d52dd18ada571cdcead2860d1a1084f1a4597c179da +size 62904 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_053.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_053.png new file mode 100644 index 0000000000000000000000000000000000000000..4f9980b6f377d556c2bb26ac1b5aaea0c9c779b3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_053.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67006757984910be5b5773dec7f4ced0dfd9393d86fd231920248d22e3534124 +size 62040 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_054.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_054.png new file mode 100644 index 0000000000000000000000000000000000000000..d3b8e8fa209857c81547ec3cab2cba28a69e4630 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_054.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8585c31816521d0347a0b3119b5f51c1926a0375b1b657f962a9b9ff5260e3f8 +size 61659 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_057.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_057.png new file mode 100644 index 0000000000000000000000000000000000000000..17f50df584a764895bd8eb3850a1b72be325f2e6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_057.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f928c83d7e0b96a40abd91729b41c9d75ad346101761b450601dfc59cebfa67 +size 60196 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_058.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_058.png new file mode 100644 index 0000000000000000000000000000000000000000..2cc643b2e44edb800353ca9871a2c979cc0872e3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_058.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6528c28986b9236070c9d6ab449c56a9664b3fd2a771c47a4a9fca6d9a418619 +size 59978 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_059.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_059.png new file mode 100644 index 0000000000000000000000000000000000000000..cb90fa49523d9e0995e7b96c44914e03d5f0df3a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_059.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38e3d31d9483739fe19e062266c78726dee50d1a6e545621edb26896d5702d40 +size 59700 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..93f611b6b352aede0313042b7beb5f8ccfd312cd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbd8121244da6d2c51065858f2fed615cd984af6544b7430b568bfaf7d4d0cdc +size 59601 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_063.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_063.png new file mode 100644 index 0000000000000000000000000000000000000000..f786d075e62e96723cf029feeef2d967e84a07af --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_063.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95c255b3553e8ac41f53f904275fb8e310261b3984406e4848b1d1691648f3d8 +size 59860 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_064.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_064.png new file mode 100644 index 0000000000000000000000000000000000000000..bcf73a1ce51bc8fcfaf94a7c3d2e199b31960382 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_064.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:422e8d3a901fd4716011598f372148c22ab49412707adc8760cadb4445577e41 +size 59917 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..b0a4016431b422a3deec0c3914af602fc06cb875 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:73df09ef2a98a578a45984a074cd0b5aebb061d3dbc1945f48417582d578a522 +size 59941 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_066.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_066.png new file mode 100644 index 0000000000000000000000000000000000000000..2b8c19bf5cc2b519843e3036e06b808d691ae272 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_066.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd48fb5b797865a4fe9110440df17935bdea8563dc24dd13c1823845318ff7e3 +size 60258 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_068.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_068.png new file mode 100644 index 0000000000000000000000000000000000000000..ca7931f35deaf89243f8455d7557f40693011c6d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_068.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf25a6cf17e351d31138c4a4ff7dae56e607af681fb0d51778e1e7881f6b0380 +size 61519 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_069.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_069.png new file mode 100644 index 0000000000000000000000000000000000000000..0d854fcc094665d1161465da4c2d4df7d6d9c2f3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_069.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:440e0768363ed40fe666a2e6535df5cb20da7e1c47d82266bdc8bd698e9b1abc +size 62338 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_070.png new file mode 100644 index 0000000000000000000000000000000000000000..aca13b9bfe8bcf6668eab9668bda87db8720d9ae --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b02129d2725abc6a372838b6a441f23c75910529005b576d77b85dffa7c22164 +size 63337 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_072.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_072.png new file mode 100644 index 0000000000000000000000000000000000000000..3190905cf92b6e4e47fd509a91fcd1a666414a7f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_072.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4e0e2492fd96c470be9ba771f1f801fa3e5cb695a96af4deb1f7731dbb8f44e +size 64632 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_073.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_073.png new file mode 100644 index 0000000000000000000000000000000000000000..10b81ca1341fb9db8049797d4dcca355b4cc20e8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_073.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa6df870ed8c6d37ab7bd86cf5c7a85430b2b52c913b7981369da605fc6b40e4 +size 64415 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_074.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_074.png new file mode 100644 index 0000000000000000000000000000000000000000..f701618c5d107482806caf8c54d3fa7615c4fd24 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_074.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b22ae6aeda969638beb503dedf2bc47cafae61944f3e6708c1ea1aba893028a +size 64234 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..ca09cdea38d76e7154d9b16aeb28dac4e98da077 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0eb674bdbbfbcb4e13f91b848a55d769b00497488a330273b1c3a6b591e844f9 +size 63422 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_077.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_077.png new file mode 100644 index 0000000000000000000000000000000000000000..f8fadd51f86d33b9699222166a08c9a042d39ed3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_077.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77ab35bc928887469b087db9ee77cefe0af7d091acea880a5790d62a7cbc948a +size 61882 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_079.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_079.png new file mode 100644 index 0000000000000000000000000000000000000000..d9d97bc57e32acddcb3c7f6c9b55c6d20ab8d9c2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_079.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a6a75b22af4c2c7139b7ce7add60e13484368a6dda2b914ad19b33d4b909e90 +size 60874 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..48e7b8381a5dc4b2a810e91917f9e9a27ab2444e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4805ea2263aae47e160873ff801a4f30049b05c772c0571f5f19864dd43dd70 +size 60695 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_081.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_081.png new file mode 100644 index 0000000000000000000000000000000000000000..2a19c564715416987af846a5b219f55dd19e5f5a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_081.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ae0d1a47ffd1951ce9679a2354d974f6d3b49db81caacaeabd9d94372ff3b99 +size 60714 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_082.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_082.png new file mode 100644 index 0000000000000000000000000000000000000000..dffa34bdf5681e5d86e601181ab8c5783f92dd4c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_082.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a39bcbdc1a3c3fcb314920e375cac98d616b71d180f3458614f93f7ca619631 +size 60534 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_083.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_083.png new file mode 100644 index 0000000000000000000000000000000000000000..1782d6f17bdc035f1fff4e19550ffb6239a5256f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_083.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a00a2942f46ba83a7a91af1d0cd0521deb616fb90e15f3dff16a1c8fbd7715ff +size 60799 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..a7cc32898f024cea303750414e5616ca62e20738 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20e9c1357046453a4e8c2395f36b9f7edf5f08bfd30c45609ff69cbe193b3417 +size 63031 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_086.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_086.png new file mode 100644 index 0000000000000000000000000000000000000000..71bdd805c2a1b411a60f94dab4a4aee06bb2c121 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_086.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1cb45ca1349cad3ca2deb6d71f0c937c1da6d403d31fbf09b690190ccbe2ac7f +size 63717 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_088.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_088.png new file mode 100644 index 0000000000000000000000000000000000000000..30abddd80bb0914135e159540aaae4b832c95871 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_088.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:910bd7f340c6bbf6670351f4d695f881e02c07885010128ca14cfef8c06fe849 +size 63945 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_089.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_089.png new file mode 100644 index 0000000000000000000000000000000000000000..b3db392ffa436c3d83cf45b3193057801e95874f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_089.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8910b5193f4ca55afc5e0d40f8e97e4f3908aa70d9d4be7475754927342a656f +size 64393 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..3a3c7144324f826402571f478db2c7605882a697 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:795a4cbe629ada12b778718a10af68116280cd7dbe14df1bf025d1fdea6e672f +size 63651 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_091.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_091.png new file mode 100644 index 0000000000000000000000000000000000000000..35e43794ac93a04d2375819fa2c85921a2bea573 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_091.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ebce624032ec7ddb5fd77b24f9c0df85f19229e17b12708c95596c26757507f2 +size 62808 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_092.png new file mode 100644 index 0000000000000000000000000000000000000000..358cfbf6901d79d1f466bfb9f9f75eb50e54c620 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9fb0667b0dcc6d529eae2ba7099a7a0fbbd7fe77ac860053a146a42deea76a49 +size 62131 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_094.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_094.png new file mode 100644 index 0000000000000000000000000000000000000000..e281c4d2ccd18046092ecbb0e0d3992c15dd22d6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_094.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57137b033bb474dd6241095439d4156b35f7a8a4e0952d163055ab48133894dd +size 60829 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..ade31bfaa1018d638df088ef9cced6b4a212f6b8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c12b9a4a9a056d8fc720be28213b1a4adc25dac9447bf4be5403eeda6764b2ee +size 59821 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_096.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_096.png new file mode 100644 index 0000000000000000000000000000000000000000..a273abba2182e31c6ebeeed1df2dac5adfb1fb42 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_096.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7267f7dd5fb04b5ab5ca94c65c6c6d24dbca17041c8d63201c6ae7d1760fb12 +size 58970 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_097.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_097.png new file mode 100644 index 0000000000000000000000000000000000000000..eaa66a56bf3558d78df0d0f8defd924690d1e462 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_097.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb56f1b2b015738d6edfa4ecaa9ddbaa04fdbc6cf5bcce7f3637a36ba716637c +size 58729 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_098.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_098.png new file mode 100644 index 0000000000000000000000000000000000000000..06b5186b9ffd10ddf0da7c388f3db9bd0d3ff729 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_098.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08de5fc668be60d87ded404e2b5606cd433a0dc34e0c9d4861b61ee2ca8cbf98 +size 58864 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_099.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_099.png new file mode 100644 index 0000000000000000000000000000000000000000..36537fb7ae2b92a352b5bc29f3c8e645f0f5353c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_099.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca73016ccf9ee6ff9d9bf81b85e599594ab2c499a53e1ccd6da6783313257d10 +size 58846 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..5fcda39c67c03aca2877293a79e7cf42e91f4428 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a0a3c9b59c50e449cde15183861329e73994e0d7dba4be4ca79bfa3351a4c19 +size 58899 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_101.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_101.png new file mode 100644 index 0000000000000000000000000000000000000000..1aa05eef83d4c946fa6ae2d6f89d27a509621b61 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_101.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b68b4f4065033e17913195f81b185a025ad31cc25305bfea2410d8f20860be24 +size 58956 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_102.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_102.png new file mode 100644 index 0000000000000000000000000000000000000000..5aaabc2dacd3026450d5c183c1620e9e550abc05 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_102.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5fe4ae0ec808dfd5e29301d13d40e650018f021de5da2215bd5f70915c071376 +size 58857 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_103.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_103.png new file mode 100644 index 0000000000000000000000000000000000000000..b675e599b4b05671288da8853f8fae0d7449659d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_103.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38c7e7e3ba2f2f0bd6dd68a6c2e5d8da169a1af4a84fafa775bc33d63ef32e16 +size 58833 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_104.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_104.png new file mode 100644 index 0000000000000000000000000000000000000000..76067733aa541d192c99b3b6dbf7b1bc3f45898e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_104.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6dd57baa48b028b4294a4d497f519e529b0d68818f52f472131920af0ef8aab5 +size 59482 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_106.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_106.png new file mode 100644 index 0000000000000000000000000000000000000000..42ba40deab7191a3ec6d6a044d4334385daf7b4b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_106.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b7c9778ccd6ba07c83d166fe5459d0096e2657eea98c2ea39a4c2f0c400529e +size 61294 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_107.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_107.png new file mode 100644 index 0000000000000000000000000000000000000000..323bba8c9df2401a9cea512e097d4480a7b30d19 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_107.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e420e8fffd229b8d753170f8bd55060c01397f6b7c843c8baa9910c9ff10ee5 +size 63402 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_108.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_108.png new file mode 100644 index 0000000000000000000000000000000000000000..514c98b285ecabfecdf18f70ec1b44644802a251 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_108.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21cfd30a759c9c530f86b6a48bdf955c04720995c56796e85a718b445e4cac83 +size 63606 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_111.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_111.png new file mode 100644 index 0000000000000000000000000000000000000000..75a96f58ef71eff20dfeafae11b4714517edbb65 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_111.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d9fad9a25ccd3e0894ca1c541dfb578f4dbd66cc9d2e5d55e77d3812a43981c +size 64637 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_112.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_112.png new file mode 100644 index 0000000000000000000000000000000000000000..2d3508e6fc6e87e565384d3efffe006d16a4e286 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_112.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b0f0c6a9a2bcfba0f21aa364d3a197cc040a727b2d98608bb8116480c2765af +size 63828 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_113.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_113.png new file mode 100644 index 0000000000000000000000000000000000000000..46f8d40c040f883c6f3351d3ba027ea4045639cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_113.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b61bee97f4d49eda6fae0f2c9b165936b95e2aee290a9689cb8f5ce9f1a20f08 +size 62505 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_114.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_114.png new file mode 100644 index 0000000000000000000000000000000000000000..bd962558282489e72b3da1ab746f9b77a6fa2036 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_114.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7cdbfa0a89ecbc9e26974095e3aeb68e6d2256b62cb08236e39cbf0fd04afa1f +size 60998 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..9631640c0d0d4ae4f02fa074d2e26461c5be1d15 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f92282ca39e3d801f7f4a7cac535d8803e00c195837fc3e87f64d559b9c25009 +size 60195 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_117.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_117.png new file mode 100644 index 0000000000000000000000000000000000000000..4108ff06a87bcf27dc79791803c9b9e66b8c6d13 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_117.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10a3aa75f069d1b55b75a8ad9aa8fa917e25aa9db08b3d1b535f2274e5903821 +size 58966 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_118.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_118.png new file mode 100644 index 0000000000000000000000000000000000000000..0a0e8e3a87cc3d7521aebc2630aa4fc4c08420e8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_118.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1df03e2201d483482b070dfde136e497a38ee2ea10661ada9812654d4ca2cda3 +size 58735 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_119.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_119.png new file mode 100644 index 0000000000000000000000000000000000000000..c7649891412ed04da4142af87bd09cc107379437 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_119.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ef6cd53ac8a85293f3778504237089a693e5bc6baab041889844f7fd839b2bf +size 58185 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..5131b8d620c4dd1fe3009a21b171e345dd9aa58c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/mosca2/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b00ede8335355b49d1f959875fc7e23a853543085de6ba312217e2128c92b543 +size 58083 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/moscaG2D/frame_059.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/moscaG2D/frame_059.png new file mode 100644 index 0000000000000000000000000000000000000000..f51d9365ade44bc61615747ed6d068f68ff24368 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/moscaG2D/frame_059.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe70e1bc82d75259a55c0d2bd9467014badbb2bfa18527fba036537e11d68b82 +size 59632 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..114d56714fdf45ff1dabcaf09fb73a32a4400d71 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25361bbd0defa0a35fc8c0d37eb0bf0424edd773e8607285d38561c3d9aadb6f +size 62591 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..d6bd14d45f60d916c1721abad6c6ff8eb97d9545 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:570fdead7f35a6d74597e12e06e8857ef820bc17f86264d3b10d6f224dba4839 +size 57975 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_010.png new file mode 100644 index 0000000000000000000000000000000000000000..600d6fdf866ab16d0335eec8cce56a97a02e61a8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e04becf332b7949b6b5ddf07dd023a533ec12afe7969098d06453f50e32e5f3e +size 63721 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..f9e305d0456b7f1537e5729004fbc6a07edfbe7b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4f28eff68c7740b63da472b37b529b31a6625d9493825fc103334564aba5dcc +size 64093 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..cf5321df1b5d29e4ef17cc9d48907ec3c4ffcd17 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:392b1e288c2e3cb8b0b70eb9932b0085f19de44c5f6a6c6878d16a8370f4b33b +size 60640 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..9bb90210bf7178e4b453f5adc2144cb3f55ed70b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43db7f98798e597dce268ff1962a3c08e5cb8b1cd097c4c491a46af1ab3fce25 +size 62679 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..ca5af390f620119e2aa8aca4448f305894898f04 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b6d29f0ef25d3c621a7880da8853bce81dd7720f387efd551977aa9b6b23bfc +size 61807 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..6fd34331b3a92f9fcbf383a5d38a88df8787fa94 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9770cd48ceafe339e5747882186cdcca750adb73932c1fee2c25cf7d5fd4435c +size 64266 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_110.png new file mode 100644 index 0000000000000000000000000000000000000000..c596bbe13b5bfb4998d57d95afdd4e6d580a77fe --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/moscaG2D/frame_110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af42e552cf25f5cfbef2d887cf26cf7404a8b124080fe73ccc053bafcb8e7639 +size 69095 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/000.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/005.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/010.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/020.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/025.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/030.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/035.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/040.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/045.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/050.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/060.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/065.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/075.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/080.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/085.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/090.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/100.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/105.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/110.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/115.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/120.png new file mode 100644 index 0000000000000000000000000000000000000000..385c98fa1ad4a8bff16a60dfb896a259f3da525d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mask_square/120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e5f8d31e14087ac6bd582b7fb8c521fe2e3654b309f8f8e804e6b1dd56ae6a8 +size 1070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..7a9e3fe47335764577d0335b4a585e504dfa05ea --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f989d583e0d985ade387f08dc926df5eda12817856b89dcfd519767dc56998b7 +size 67895 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..a39a43ed04e67f728ab7dab553e76e42fa9870b9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:452a0d066899ba38ff7e123fc0a3bd3bbb180767c21f6b2db65e7bd38e1bc218 +size 57768 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..cd23a1f71ef9769c06663e7fe843a2af3afcb4fa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96164e1edb916d0521a44360803b104adfaac0ad00465eeb76b959420557205e +size 56450 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_010.png new file mode 100644 index 0000000000000000000000000000000000000000..7d65d67c2650745c34e78d1f1cf1fc652181255e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2db9c8e8b57ffc9cef55171a734f04f21085e9b4820155d60d4487c7d8bdbc8e +size 55898 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_015.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_015.png new file mode 100644 index 0000000000000000000000000000000000000000..ed1379efa4d49b914ba422c8036447bc261db58d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_015.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35ba21333ec3b198bfbd8d8af27dfcbe7fed8ced336edd56c603fa82ccb2dd78 +size 55001 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..dbf5fec2054b3ab8d4d716b5aac15a41cfd80705 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57b9eb1dce6dd8281773e275b02f480a60a5db06eb28fd2a87ca8e5f31a4022d +size 56518 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_025.png new file mode 100644 index 0000000000000000000000000000000000000000..54b2c3f27ff74c27ee6329f89fcc3784d07c6876 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b81df9f82787fd4b8f9d39f59725208541062e5c2eb7e353d9ba060d1fbb9d7 +size 58307 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..9214045f43cfdc4e0be9893ec481dfe189b93b8b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9d5b9bdd9ca8d0a3c20236c2cf5008faa98a0532bd43388735ea48c71cc8d40 +size 61757 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..db661e67f676eaa9b7658450322026deaf94b2bd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c26f4558234618690359bb50540390ae881bb9f58ef238b7662a2dbb2b74d2a +size 60253 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..5ceddb3a4bd5340108a3dbbd6e98005e97479d88 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bdd38c7937f63c392a3c18fc483a57b66b5b823f3f585217ff7757a7366f065 +size 60097 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..dca86ef697b777f3504a73745c64cb89d5700585 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52a88afc4310ac8f6c349b986b17487998d9e0f7125cd377a1b44b50899bb44f +size 64668 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_055.png new file mode 100644 index 0000000000000000000000000000000000000000..d903beb7ae7b8277944a4e0c328733e6b85e4b77 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f9454f6d0caff7104ecc4e9443a18cc530a044da766fc95011f758134cae5a0 +size 65140 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..6744f9b1e9cdc4f1df5f877fe83b2ca92cdb38a7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f2f6367359dd2e19e88b1c550c6602f48a0106c7960f35fb8434d4c70399227 +size 65413 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..767d781349d6ce643c96e269aa39eebdd239a7b8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e4341662476fcb36778954c22dafdbb24dcc3f3876ce5fdc0429821e43315ba +size 63866 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_070.png new file mode 100644 index 0000000000000000000000000000000000000000..7509cdee1e08dcafa3072a2c5bcef4d9aa7dcc2e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3524d3fc1161cc3531a778a2985bcb36eaf5e6d98be951d3f0a1f1ac7948e3ec +size 61741 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..5a6f601f8ebdb4e48c2b133b8536a65c5cbea70d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52e69a90b1c1c35dc4b7871af75d053500957b33d513a7eb4498e171b4c7bbc1 +size 56770 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..e5a57bb8581b44917093ab27bfd97a26327f4d87 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7faecd4e844bd9721a63864f7e1706d4c063086b6a5c6fa2e31bf94d0857814f +size 59335 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..503c53247baf9b204d609473e596d27028d067df --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c195604d2a91a4d0aa4c0bb6db86972413a935e0039845892273bb3eb5106ec6 +size 55610 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..3fb679599443f918aa3282bd976fc9dfd9f26763 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:095344a8e2546cb77e2b7eff8e5c573b81305733fd98a0ae4e8e20701adee903 +size 57900 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..b533b753c1e704627330de2af577d9d7e40e0cef --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26226183f6280596a03f72041bf923c17d0e30a63a0ea5f02936fef235581efd +size 57110 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..f38bac7d4e58e3a4edc8905ab71ec81ecba9bdd9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c87782c9955e98400cf584288ed8b00a390639a648c912a8bb762a26362c0663 +size 57973 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..4b10622feabd7860ae7baaad524f459572769223 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:700fc317c83c8faa34543272ff91c6c7fbeef055cc05972da28c90842ba2d498 +size 58868 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_110.png new file mode 100644 index 0000000000000000000000000000000000000000..9c396ff45825907fa6ea9969abd964936c343639 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d33906f138dc9e0162f5fdae545d9c9aea9c7c9f5eb440a5ca91e90c78855164 +size 62405 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..9232b612cbd48649406f2495d07528d4d3fabbe7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5715df6996a900a5e810af70df8035d6cf1f57af2fa842f07ebb4d2aebe0aac7 +size 61172 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..18b2f549e9b6a76659180ca98d76ed33753df97f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/mosca2/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04ae119c4c9cf32ac145325a0f5afa204262404416e0f82ef41e783d5ec6ab89 +size 62923 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..36e8447b4c87b5bb3ac93294f201a79de888259e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d820ad69a4c062f9dfe68ffcd9aa94bcac8e142d8ebce9893dcfe9bb8dbf8ae +size 94734 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..ecb7ae0fcc15ae1257b82c8e42215f88c2323daa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef56dad846ed4cf08a899fef4bab0293881a1c40e5e2e52ee27fa0135622ab23 +size 67214 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..86777f9dc0ec24ab83a2cb1ca2aa1f253ae95050 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f17a7afe8d9989d34a62bc52dd0d75e95c5998324e6480c4b41b0a12ef83dae +size 61766 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_010.png new file mode 100644 index 0000000000000000000000000000000000000000..589b1b3bda4f8030801bcc7c934532fb29f40acb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22f2184a914e2a83ad9ac578d4c9b0eb11b36ca8e1da87489d85dc364b539c63 +size 63240 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_015.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_015.png new file mode 100644 index 0000000000000000000000000000000000000000..7f46eb22c83bd9eb069145935cd2637af254c3e7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_015.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17e7ba792a07a634cc37545d14e27816a3382e3ca2540161b9274b60f4028af1 +size 58106 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..d0eecd87a38a116c840e85cebf5389c0fc409a6a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3738f3c52c79793f93c8a51f859caeee90f4dbf87c582c546e631f2a7627ae47 +size 63317 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_025.png new file mode 100644 index 0000000000000000000000000000000000000000..f3b7a98c262c30585512f87f6386e38c8c284bef --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72393af618e0ca96a7134d58f1a6fcb73e3ab154898e36b603302118fac3b936 +size 67430 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..91733099039f6b2a8881b81e2732942576f13a81 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd98c44bc2342b3cdf83c44a031bd0f9970aed99b142a6e6d9af91a1eac91fe9 +size 64746 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..10ef2447c91546d59d83530291e623c1a9407e07 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a8257a7f686f61ddb45be986cddd933f19b366fe4f9a957dd7b00ba1dfc184f +size 63685 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..1042aa22758a6986fd45d6fea87f6f30ec1df19c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c361b0cec41c93bf6c41752667791240f0c91734c26f4f0ae7dad1573e78b32 +size 61966 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_045.png new file mode 100644 index 0000000000000000000000000000000000000000..d03f3c575b40560fd021a8b14ca3c1d37f42a93c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6c02d2c6e95dfb3abfd1af6d2fe05fd22c8920125421c5e467bff469fa0446d +size 60669 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..4c03af0ad2fcc528e05778360d4fa5ea8a804bf3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7379b49719fd96f569faeb5c2d21cdc65766dd1c32a1c01cce5b48ca205c2a8f +size 66075 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_055.png new file mode 100644 index 0000000000000000000000000000000000000000..b016b201d142b1865a9f65333fcfc76887e566fd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:652fba9ffecdb2a9c5f46bfb9722d02bc39b5e659ff972157810758cc48588b0 +size 61291 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..f56a0ea90f9ef6381abb7eaabe0df182319c38a7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f346332b210f818fb3c9869da4c0aafe5f7af036ad1bfbbdf7553ec3a9bd2d58 +size 64979 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..7a8852d894c38d11183fe7f850929c2c6e0e92f1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aba6a9c21b519d2928bb9f47b7eff24792fa70aa1f1c03b019d2530bf124e0ae +size 64982 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_070.png new file mode 100644 index 0000000000000000000000000000000000000000..9044de39c6efa07d5c24b30d4ea6f15ad2d945c4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c86cb55f9503e1c548ef4490bbfe95d45f6fd4d964345903d3a22bc857c99b3 +size 61172 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..14b89ea3b77035bc31a26aeb155d09fb896e5bd0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:492cfa35b3040994c3bcd453512995223610f7a45bd73bae8b0039f2f4450208 +size 57339 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..80621cf365a109afde377e2621b3ccf83292d44d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a78462aa29920c0a7008805ffd518e35906b973b1286dfa1753edde164ee8e0 +size 58820 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..8c88c6a44d2cdf97df3ad4c1a46716c541303a4b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e82c3ba6c1285ad5ff67a304fa22c8b5b7caa73660724e1b95ff663f65d0ba48 +size 58371 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..a3ff0026baebab618661d1d41b69a85d8129506c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d8f9cfedb62e2bc2ff1896ae1e975bcb15acb60a05184e587f6d47e7ac264490 +size 62186 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..00d8051d75736bdc0105871efd54b03e9b800b9e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d11b29cb22af8794c03c98d218c0613b147eb993092026bdeb9cbff842c88e0c +size 61946 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..f46bafdadc1fae6b56339683ec4d29890be3f296 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85a005054cd8b693ff8b752bd270a2e8a5e5629006720d3a8cd583198e3ace21 +size 62111 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_110.png new file mode 100644 index 0000000000000000000000000000000000000000..07a63fca50a1158185bd6d7bc389dbbbee8c0cab --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dabbd497e49a5a300a5471181dfcd36a04364ca3ec3eec091171831a2a9e7cd2 +size 67439 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..c105144f6f2b4e66464579f0690a0f88c6913abb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d88224fc6df18b0a5597ba21e24125f678d7d45ca23bb87feb2a176c389bf6e8 +size 69103 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..1ff71abbcc05db80132d6e95d3780c5535b7e0e1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/moscaG2D/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e27ad1587647f8835daf0cd08d2ca83e51b0450d27c6399a323510755efac22 +size 68471 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/000.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/005.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/010.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/020.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/025.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/040.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/050.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/055.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/060.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/065.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/070.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/075.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/080.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/095.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/100.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/105.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/110.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/115.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/120.png new file mode 100644 index 0000000000000000000000000000000000000000..3485fe63d11e67010176247c53b91c00163d54bf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mask_square/120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:293ae574c63b39815ff222b90245ebf6f6a47ee28505f5d039dcdb10c89fe9f9 +size 1066 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..f739e4fec02f870792bf88c63a26fcf965110e72 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de7bdbb50a302fa45fd5ece762af4adc0515347336b02a957aa94bb5da203778 +size 62591 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..5f657d7c94d51a3265dfaa25ec6c1d6ad89cff8b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:908f041085a44b5a03ae1b72c1bad55b4e6cd5b36860d3192cb50c9df63d4e57 +size 57877 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..b11199c1d84db7e8ad4b1839723fe9b05358a3fd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9eeccceac98caa72d5722c1e81ba6d5584c29067d1661e60e94d20c3cb9d58b5 +size 53536 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_010.png new file mode 100644 index 0000000000000000000000000000000000000000..601d21490f70dab884037fc0ea5dcb5a6885eae8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d982fde68382ae46302deb2560f643cbe33bc1f40205ced01b07d5fe5b4fa82 +size 52811 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_015.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_015.png new file mode 100644 index 0000000000000000000000000000000000000000..7b64777871729221a4f2f268f1b90bd08a6303e1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_015.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70b5c965d3c22ef4665d92fdb4be5e20dd301ded3838895b4a7376e6a0870bc0 +size 52726 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..b49f6306ee07c52cb4cc0e9dfc2dde355a22c46e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2545211fde304d4e9242c13418a149c78670100d52861d317f2a9bce787c4abf +size 51903 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_025.png new file mode 100644 index 0000000000000000000000000000000000000000..47f2bb8e68dc87bf7e5fb5c5b06ddcb5abd45e0c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74d6ec546c4dcaa1461637f732f91a656fb118eb23684550d0a2ad612c156c70 +size 50337 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..85298dbe0dc3c6a0c31e2c65e7b5bd7fff5c7277 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:133cdf72d16eaf0df47171d213d87a4c108fcf7dea5dda1b891a5a4131f625fa +size 51726 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..73e82733fc208e5d289b32b7a8d01a88dc3fa977 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e75336c377ff7cfa4e4ba84b1edca2f4e3434d2e13284e8d35793890248b894c +size 54953 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..137e85e630077032dcf1b0b3cbdc26962fe484a7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c3b3459ef7e440d50e3ac03c971e67290415878b101feda26e302375dd98a63 +size 55101 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_045.png new file mode 100644 index 0000000000000000000000000000000000000000..f667b6d1f1462d2b5653561984224e4b48bcff1c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9b52d3e2107339d6cc977ab83d3c21b6b5026f42122c10cb10653971d76e5e6 +size 53667 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..bd7d4eec014dd8ae9567f1ecdb06325fcadd1729 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a824a2a606375581e5eba4c9b191f288af767c76a4487635c80e02c7d89ec6e +size 55826 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..cc928b89ed84751e6ad4dbd95b85b36af2bdcb71 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f9ec8cd867b52998d8df3d47c125aee868ce55490329783df5f254266184c1d +size 60404 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..bb418d36f33dd558350f526be4c47ff80b65dbc1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9606b5383466a933636de5cd8b7d610460de9e6a459cd1bc5662c8eac29cbd73 +size 58615 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_070.png new file mode 100644 index 0000000000000000000000000000000000000000..67386232db9e6c6302c793f8d9bd28bf01c5eacc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e23ec32be3c52629c50f6caabc2f5e5957762df0dd824935d4e4e454e86aa96 +size 57671 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..0ddf95b04255e4824c205a875020f7e430554490 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:031368eb3cf2b3ae52a18dd0498862b02a6f531c0541e10813839b738a343eec +size 56107 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..4faab08a1e65d52377e0b6dbf5f80e509b1903cf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f61852c78e00a4c02be29f1b59f024f0082541e3ad54c505148013dfe16d875 +size 58187 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..3550972c887eac213a9d686ef4a83e55f5721e6c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67137bb9dc70f1369ca488af57129712d42126fce38097a6567cda8ad43b89ad +size 51227 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..b91dd3aabb8842adbfb8efe2ee0e2cfe2a4639e9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:598ab229e4cdacc9d4774ac29b3cfc9194bc787785a0354a859cab1e3720132e +size 51475 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..2d5df2a40b286c7451601e0c0105867f49bb6968 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2fa2d0c54f5e2a276f1d7b4cbcb9c86639b822eb4d196abc802fd0c1d3adeb4 +size 52911 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..59295814637bf8d6a9b04ac023ac43987f6cd3e4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e0a511699e2569f0e5da8740edda74352d00c9478129c50e6432cac87900a86 +size 52715 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_110.png new file mode 100644 index 0000000000000000000000000000000000000000..34a53806195252b9a3eba4e7b88022d355298832 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02fd6fdf2ae6b7ac1bce9893a7dd3ebc2e8b017354798c5f56b3fc5f66b641e3 +size 54495 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..06875762306fca0f9f4ae1b73738e4cf174e4cdc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/mosca2/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99a7683ccd7de02856160c35c2fcb66d6417f44879974414eea380ace82d71f7 +size 58896 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..d4050073c456032e983e8a4158f4dc446d2d4f2f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:377eb69236a59f4bdcae1dcd7e054fe74c4de5cb506a24af8dd3c5b57a547339 +size 89773 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..e5dc07197b8ee943c862be80a6cb8230f8ea9122 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a241f0cef0e58d2b465260c148c0df48e25ff762e35e18b30faf39b483a5b74 +size 62025 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..9e18c4511fc825f73b83e179d12b98d91b01f2a4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8e6d26e3d0b4bcfce20e041c35ef61803826940c97a16fdf4e493d7bce2994c +size 54506 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_010.png new file mode 100644 index 0000000000000000000000000000000000000000..9733efb79364c61851b7a968b865112cf946a80f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77eca7c4331ed7f55d49f4babfcb469d408fbaed6b2b1d274556248cfb8908eb +size 60516 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_015.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_015.png new file mode 100644 index 0000000000000000000000000000000000000000..ef41bb4df64f0e5e2a7c23014e15ca1b1a70c885 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_015.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:963504e763615cba5c8e8be6b6f36a9b42d66169238123b429b445311e7a8816 +size 59699 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..b38f5f46e94511c490cd95b95ba2058ed7f3e62c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9929fa3ba23de0bc577be3e75118317aa919905553c706bb93aa05b5788ff7ec +size 57387 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_025.png new file mode 100644 index 0000000000000000000000000000000000000000..ef2671a9b54d0ca0e21f5a292d628bb63c41e6eb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0204c4fd7df2157390b757716daff5cca8f3a5c5146f48179719fefeea17f4ce +size 53273 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..131349521cfd0b8c33a55b5e3b4644f2203b9fcf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f88cf112e67fb20d33cea685db3b8d061f1cbb9674d72ca211882227a5edf27 +size 53807 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..e5bbe62c710363322652b6dddce7b229876df963 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:431ff60e473fd6d95aa4e86ca4bccefd70c20166c661dcd055330231d58e8260 +size 58683 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..daeb791405276de199c62c444b148cddac37d2ba --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8815552a21f2b12504254a86017ce383cf023cd90743ff3d860098713efa942b +size 59893 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_045.png new file mode 100644 index 0000000000000000000000000000000000000000..10a5f5bda03e01a1f7f8729e9696b192020ba00e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e0b5e56fe9819e89fbc18d4e15854d31275d44ed4cb06633e8de504e5438e46 +size 58315 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..a674b499b3c104dbb9c3b535d3643ee84adc94d8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af6a6ef934130f5c7b87e972686c485dc83172dd697fc716f8a97fccacb3d1cb +size 60070 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_055.png new file mode 100644 index 0000000000000000000000000000000000000000..43cdc753d34a283ee5a4d3936c90b9e498f2b4ad --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:080caf4879dc95883a500887bd53f13344be190c2f44fbb6237cc0dae68e0da4 +size 57922 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..9c57630e38108ce5a488ecb080672caf97759a4c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95d0eb62d3ac990d213cf4accaa3d670d6c8d7688432d20a5d1a27d7e623489f +size 62855 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..fd4bc612d0c8243eff53942dccd395e409461af2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:790a2f805a4136a7f4f2b537f270097ef718fa2a95cfc931676140a749d87f07 +size 62143 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_070.png new file mode 100644 index 0000000000000000000000000000000000000000..603975a53e5ed14cf342bdbd59c5fccefac34448 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53ea9f7155426965f594c4c4c5189893decadc5f8523b224e73fd90eb7642573 +size 57633 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..f4cb0d15b6176c6a6cd020fd06e496173750d38f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d21b22f8c6089d70033be15fc5feb434c6d47350982e9349c785880c0e1c534 +size 56901 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..77cac20d6382ee37813933409ce87167804ca613 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf41dd6f22d1e46bdea242a9a17381d37218834e06a888e0a36bced94b30258e +size 58244 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..5a2601504882216b4627bf65aba450d9a4fbb404 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b87f4c71848b5327bce5a4b9138bd157c9064fc689622cb11ae0d57248a7c004 +size 54408 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..e398d9b4c7e4a4b7f2b6f0cef57c2d63d25a05bb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd5d5b2c782b265cc824c886f2ca604b081779fddc120ec39766ec093ed39f66 +size 57444 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..3d4dc2c2dafdb6895dcca9e297c0404d41ffbe97 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:262b33b7c8e272ac09993aab3932a0513ebdfca40cbf1825e2ad6f19883aacae +size 57838 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..240633995ac9e547a744510887f48b15ed797cbc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a7d910c9af2139d8b9522d6eae3e6eebad43b587152512b61ba5f5b9026e907 +size 56454 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_110.png new file mode 100644 index 0000000000000000000000000000000000000000..0aac721aeb9c25a4d9553a92e32085e4110646a2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0e659b9705534db6e2c849b6068564b3c570ea291144e8d4478e076021f99f0 +size 61410 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..d66ca629483de28e75df0c2e0158918793959b84 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4ca252823a11fedb37638cb2467d9a2d9d07855cbafcd0b7b32587bcb7318f1 +size 61688 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..b5eaa2ef4e34ee8cbafeacbbe885937e495b7ee7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/moscaG2D/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4afcd16d33aa07c3c251dbe1f3932b64e3566c80f018be56d895931b8fb2157 +size 63335 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_010.png new file mode 100644 index 0000000000000000000000000000000000000000..17a5a63b06bdd235388b6882296a9c9a857524a3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3ea9ccc94ba534dcca418bfb7c48c87a774193739f1a12880da35fbec42697f +size 53867 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_045.png new file mode 100644 index 0000000000000000000000000000000000000000..bef2f1dfdd0366909ce0efd1665c82be3bfb23fb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:642b5016df89ea42d40a7179ec519b118735cf9731f11e97b981bfc63119182e +size 54172 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..da3e5edf6e4bb2372e88ed164a95aaad3920b28f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4b6624979c06b3c45f6237440e55ed9458f7cda2c8fb7860e5b09d421f2a28e +size 51411 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..c85bb490e903d7b56d1d7cf51a2723accd0b9544 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ada437217632d51c3b3f008a9038122e560d89d334d13926d668d354feb763bf +size 51198 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..d5faf84cadf7e958f1eae8e29a8548362ecbf8bd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/moscaG2D/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60a202d01ad8589a3029a893e36d60b8dc0c53d2e26c07e4c6cab1a6fb502a65 +size 49857 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/000.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/010.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/015.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/015.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/015.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/020.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/025.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/030.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/035.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/040.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/045.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/050.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/055.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/060.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/065.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/070.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/075.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/080.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/085.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/090.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/095.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/100.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/105.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/110.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/115.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/120.png new file mode 100644 index 0000000000000000000000000000000000000000..0294f3c9d54f882efa6c3bf0bae0eb2c026d74cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mask_square/120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16176f786d1a76a4c23ac9927959bfcbde8b11ab54aa528c5457c7c7ba58e98a +size 1088 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..05196f67a7487cdc884bc7ddbdc3bfdede56aa39 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb806a79c8a664c1285739457812f903c8c6fbc4a005427593d91f468a43d4c8 +size 57743 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_025.png new file mode 100644 index 0000000000000000000000000000000000000000..dfad0b871b3aabd043eee7e2c758b97ddb2e57a6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:55946bda9e1abc791d44eb68d45e29813b9a045d696d48681b77646116f9688f +size 54981 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..00f423f42bea4aeb4169c31eec2b4a8977765c95 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c58d443828d54a568ca7d10517eaa62311fc91c661b5971f3d2d1c58fa2b5981 +size 50170 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..ee2db32e5696ca3286de32c7984ff663d31c8bab --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:797cb7271c62323873735fe02caaea6016d0ed39d023fd57deae6d1dbc751751 +size 51271 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_055.png new file mode 100644 index 0000000000000000000000000000000000000000..379aefca4e2f8b6cbba5e9fc92d1fb5d8e926cf0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:49a437d9ffbca81a70f0ad41b8f5f7b5e7af0a49b0a4b63e4ba30aecd28ff0ad +size 55888 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..9998ea500f07a3450b67dcf84b11d873c74cdf69 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a28208f4ff951517fd6716ef731ee95dcfa593eadc29d51e315bc148aeaa2e36 +size 58068 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..5f9c1a70dd10f8608e155a1adb12265b9465ea4c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef8a8aea7e694d278426d698a6b8f28e8755799d599e49c9e25111ec0da0845b +size 59296 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..5285723dd36cd3d46d8231085f7415001f4f4306 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a5981aa6a230dc2e26594f397b01230a8ef66cda75cd52f699aee617bef4826 +size 60304 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..0547207b4030a61b58f3eee3954c2f111bc52e88 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:221a66e88377b0c2b65d8d0c236523ca6986f208c921c6f915af450f94102163 +size 58909 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..475dcc8df693032c712c914cc4ea04ae26ff85fa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da52b0fd60a4e25b5352de68ebdc0c36348dadf23bc8e4275dddf77973a88d20 +size 58636 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..430cf3df679e296cdcf24dda5c9d8a905632e87c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a9736ad1284341cc0ebdb4bb3591ea7343dd66ede2de050c61eddd92ec096a0a +size 57890 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..77b10469a309f4a64adba406d5f93ee8f100ddb9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cba9bb6eb904cbcc31dfa6086504d66d436f3f5c9cb14bb7f0eb64318d69b1ea +size 58831 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..b545edfe036b36850da43cd834914da73ee7be99 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/mosca2/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c325879a4791d09ab0477a0f19874e679ca77e26e9b3f6c2d37492ba88e3d34 +size 60786 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/moscaG2D/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/moscaG2D/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..5460b1960af1f477c156aa852ab1ab55c0389031 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/moscaG2D/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f84df51795a97a8bb1f9127ca194d8c2bf687900314b02587f98c20214dc57a0 +size 73732 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/moscaG2D/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/moscaG2D/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..8bbd0764a5ae303afb3f8f7fa017474cc8f02c3b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/moscaG2D/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5399b77aa6c20d044b5b0c5fe8cc4626cde1410169d860d631f89aeb4cce33e0 +size 50263 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/000.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/015.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/015.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/015.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/020.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/040.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/045.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/055.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/085.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/095.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/105.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/110.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/120.png new file mode 100644 index 0000000000000000000000000000000000000000..5095b0a1313e109fef68e54f7f891ed199b7df86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mask_square/120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead6c5f936eebf8b1c3d2376804ad622ab862d58a2d79db16be0ca3264311ba +size 1108 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..d8bed72da2811f338b9bf7fd75a6efdf138a7232 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78421fd934c5d6170276194cfe8d6c0917117d27a439f5b4b4ce1ad67d9a2e05 +size 62434 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..bde901dd1278ffccbcabe60966ee24e53d9940df --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95422f5bafec2ed81808976539fb442ade68666416225b5d03dd1d6030b80f4d +size 66362 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..86832fa934f06797209dae4c417abdd77535bd13 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95c11d3314b3300953cf5195d9fe4b0e123ba660041cc3a3b10e30f6e818b14c +size 62941 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..b972af0db5044205c706cbbe7172357a53bb4bee --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79177ac762b688c8a81befad9bf3d6be1c5d4c364caaf0e4fbb58bf4fd893de0 +size 64956 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..fd9323264c5dab4d81e3e79cde2ea8e4ae646212 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/mosca2/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ea1b71d2d7f11ad092a38e9c3c20b986e64ae2c88b8fb07fdda454dd1780d39 +size 66248 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/moscaG2D/frame_110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/moscaG2D/frame_110.png new file mode 100644 index 0000000000000000000000000000000000000000..b8b2154490a74bb1c4b255483b7e03dd821cdd39 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/moscaG2D/frame_110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c4802cc50e5418e64d5734131ab0f4df359cd3e4582ecfac35fa5ee4149c843 +size 59686 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/005.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/012.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/012.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/012.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/018.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/018.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/018.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/023.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/024.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/026.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/026.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/026.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/034.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/034.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/034.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/041.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/041.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/041.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/047.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/047.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/047.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/055.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/065.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/076.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/076.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/076.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/077.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/077.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/077.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/090.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/092.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/100.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/103.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/103.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mask_square/103.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..5ab849c76b92cfaa679eaad25d2beb80f92ad3d7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e92f34ba0f4de606e8d346933258a6e67313faadc74635073e6775d5ae60a50c +size 258756 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_001.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_001.png new file mode 100644 index 0000000000000000000000000000000000000000..09cc4a857f7861e81938d881a7857be72263553c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_001.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4db409b4d246112da78bbb29bcb53ceefa47d325447d0efaada1f84b44b8435 +size 54705 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_003.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_003.png new file mode 100644 index 0000000000000000000000000000000000000000..4795727116382226bcf9c2a450f8fe60a7ee4c01 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_003.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b72f04c1c7897a99d24bc401dd80653bc378029ae3079b35841e292dee3f8451 +size 52999 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_004.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_004.png new file mode 100644 index 0000000000000000000000000000000000000000..5f3a5ac8903b117e7874267ca1a4eabdbb52dcbe --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_004.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93002564137c32a3edd80ecf15eb29ae305b61696824afcae8df16ae6104d353 +size 53304 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_006.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_006.png new file mode 100644 index 0000000000000000000000000000000000000000..ddd26624b6848997cf4982c79c902a3db75b92b8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_006.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d078bd5f6e445e04ed6c5781b0ef722c56ae9e74fb645fd6c608927f2d47bcf +size 54408 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_007.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_007.png new file mode 100644 index 0000000000000000000000000000000000000000..ded18afcd57d9202eb50e267ce14975e6308988b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_007.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec6a9a1a459307e22a7e0ce425b17cd05e7f745d1dbbf1453411a5bae4d37ae1 +size 55817 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_008.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_008.png new file mode 100644 index 0000000000000000000000000000000000000000..508df8f1738f761e63805a7050adab4c23f4f7f8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_008.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45ae4bf0bb4561dd678bb0e4d33324d7c376b702ea7ba13320076aa20d4ad64d +size 57390 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_011.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_011.png new file mode 100644 index 0000000000000000000000000000000000000000..ccf1d23b51aec3d964be4c9d532e2a5c7e272d6b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_011.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2625c3dc7e35408ea44477877c39db2bf1f314a49d5cd2623ee020f937675886 +size 61256 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_013.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_013.png new file mode 100644 index 0000000000000000000000000000000000000000..3887f5a132d900296e009fb3aa99c7ccfd34e7a2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_013.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4d976306748ba7c0462779b8e2571b43dba52e12a166a292f1a2848b32004fc +size 63363 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_014.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_014.png new file mode 100644 index 0000000000000000000000000000000000000000..895ee7bfab993aa61bfceb4424885a68568f290c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_014.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a1bafe1768aa9903a45cbc75e0756945b98d0aad7c439ce74f96f611f29107e +size 64280 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_017.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_017.png new file mode 100644 index 0000000000000000000000000000000000000000..81763c6a6bdfbfc1edbbd41c938fb7bdad4f2908 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_017.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c23c2146fbc19cf68520ca8314cdb319ed3205983d4d989f9366248a31232962 +size 66935 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_018.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_018.png new file mode 100644 index 0000000000000000000000000000000000000000..4c5765ddaff7eb570bb24aef448d7bd6c2eb3b1a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_018.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:40f79aa4cab3d7ed17ced9efb6f6f3d41fd8db79f4a58a63f2ea5bb32fdd8559 +size 67811 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..2795175af9ad0594775a9c4c59fa4b507c2ae023 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bda350df95435ac00fd54720f1ccf60418dc3f923222d056e6db2a8d2eab3797 +size 69579 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_021.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_021.png new file mode 100644 index 0000000000000000000000000000000000000000..13b703f97277deec02569008d13c5ec7c2c7b449 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_021.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c3af223abb0d3a6de536ec3e64827c81731c0e13ddc8d34607000e0689237f70 +size 70076 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_022.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_022.png new file mode 100644 index 0000000000000000000000000000000000000000..cfc3efd0627db856af5705b948c5e8e356934de2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_022.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:40f58cce840b5dd003d877c3f24298935e652951608fc752c2fec7ca216a22ba +size 70472 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_023.png new file mode 100644 index 0000000000000000000000000000000000000000..f0bcd916b748670ea887b106495180e39e535faf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:900829ce92f22baff17a724060381e70c74660586d56a2020ca76707315d042c +size 70259 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_024.png new file mode 100644 index 0000000000000000000000000000000000000000..027a02df52b12387598ef56203c3be367bacd436 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc5a6d14fc30f268e04858d4be44a4d10cf5f538145db1d16df3a99d6651a482 +size 69241 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_025.png new file mode 100644 index 0000000000000000000000000000000000000000..ac4c9ddac89072e81fbb22092f07b381778b0e30 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:229416c9693bb9915f5b34a99530aee035732d75d8b04a4c6b3b09e4cf50cc36 +size 68146 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_027.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_027.png new file mode 100644 index 0000000000000000000000000000000000000000..c7677cf70d4a8967e0f83e9d8e88cbb37e3b0851 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_027.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:411c861cd997b374bf38709a2c5009f22a812f391a2c73179790312a24b2b173 +size 67009 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_028.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_028.png new file mode 100644 index 0000000000000000000000000000000000000000..d8bf014b2101813b2c95bf9aa71c00806413c30f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_028.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8fe0030f61321b6b40becbd4c7ae889b66a0490daaabefc458221593aa12576 +size 66293 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_029.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_029.png new file mode 100644 index 0000000000000000000000000000000000000000..ad81b3246c74db318339f674b35241e67382b1d9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_029.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:574606528bc9c92e2db52517c8a091649639c60f44347ee758f7d80a73e5caa6 +size 64931 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..771caf5b03fddd8b204fc1d6fb68234596567b8b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d94167298de4ea1ac091051c086c9faee8e1aa9e9e25812b1f75a55700326a9e +size 64044 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_032.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_032.png new file mode 100644 index 0000000000000000000000000000000000000000..53f23a3801b76b87a1fb0142c16c51f2350e1641 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_032.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c853ccbeaef14af24b50699982f743273945bd12b436276046ea7dca549fef6 +size 64665 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..bf8f3c4a8faca355676867848d1cc8459b9697fd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5398107a604ce883b960207410dcb30237b0482652103c7368cbf0978bde6e7e +size 65192 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_036.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_036.png new file mode 100644 index 0000000000000000000000000000000000000000..cdb01287877d343519aa154c041d77961088f45f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_036.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f0395fc1d3036a758fd3c8b45327fd508615eb3dae17976976caf3d4100f813d +size 65202 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_037.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_037.png new file mode 100644 index 0000000000000000000000000000000000000000..40640b5817b3d202ef3b1200eb5b38dda34c317a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_037.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b65f0defa9f5b9d3dfb766e7bcbd15e98905670c5388f16ca3527e652fa48d8 +size 65285 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_038.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_038.png new file mode 100644 index 0000000000000000000000000000000000000000..e7969aa36e09221d83b37f74556e27eccb5b5b71 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_038.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86796ac9220832e648e7de34c15ffc2eb5be512e4441767097381b162aab8816 +size 63863 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_039.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_039.png new file mode 100644 index 0000000000000000000000000000000000000000..fd222a81de9174bb07e4a21474e770592acc4196 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_039.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91919eacfd41a5b5a8b25116d35bc2bb4d107d0a574fd8e8c33b8e55cda66ca3 +size 64117 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..55bf9d779cb664de917d45dbe0097e45f9ebf2b5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:984f6510ffcae37eea529e78ef6b22bb93070a52f0b101c01571581c902d7a6a +size 63507 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_041.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_041.png new file mode 100644 index 0000000000000000000000000000000000000000..fd8d9f698dfecde492a16cfccff5bbe4c33eebef --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_041.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cad59f56080adfb5aa88219a88167c03316fa2106b5843fb37fe22525f9c6e61 +size 62754 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_042.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_042.png new file mode 100644 index 0000000000000000000000000000000000000000..8e35f3425056e9526be44b7657b03bfe58fad33c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_042.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:676426324d093c616daa7ec141d0b9b245559d2a8c2df6e1eeb77885b7cc4391 +size 62354 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_043.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_043.png new file mode 100644 index 0000000000000000000000000000000000000000..c3f20bcfd748f2920ec1a42e5939fe8fff90f1fb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_043.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ab747c3d1183b8db8d74c08a88326b67aeadee79339adb07974f9c8eaf08d4a +size 61744 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_044.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_044.png new file mode 100644 index 0000000000000000000000000000000000000000..8b09a0d4bc62a76f60139277d4fb285178594ed9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_044.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52f526302652809a09ded2c7c854682a5bea7fc67e615d4d7cfaa17f8d3ed00b +size 61069 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_045.png new file mode 100644 index 0000000000000000000000000000000000000000..e65ae9aa53e0bbc46bcb56c4761972a83f631626 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:edda3fa99faff7e3b373abadd6f7bbbd2deb025cfa7a9974501666c0ea29d243 +size 59879 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_046.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_046.png new file mode 100644 index 0000000000000000000000000000000000000000..970358489ebff5de984af78832434f856f440cba --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_046.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01bd09746b6fed02726109fcebaa6b227b069d8f2ce60f43b2526a1e85ab1c9a +size 58549 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_047.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_047.png new file mode 100644 index 0000000000000000000000000000000000000000..d74b9951048f6d5f417106d2fdb804a1765e2c8d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_047.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b85098461700b1b04430f19e359c7f188375df6cdf24e97de5767df79d8e17a +size 56642 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_048.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_048.png new file mode 100644 index 0000000000000000000000000000000000000000..1b3211f66b4046f1d7675331d3d8b6d7eab7e4b3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_048.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fc70f53c88481985d1c6fcba921fe8a20f9da44d85d0ecfd0a2d2d8432fdc6a0 +size 55260 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_049.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_049.png new file mode 100644 index 0000000000000000000000000000000000000000..d72be6d8d7717644a873d19833c99340394ec4f7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_049.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c9b0785ad70d0ef9ad00c58584a446ba70b41b74ecae6b1500fc4c65ceffcbd +size 55690 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..bf284a95b9ac14a851ee26712691e65355c4a5ef --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83f297e5f5ccbd9da75f52fd9e34261841e8e58df166b084959e57da7939751f +size 56258 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_052.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_052.png new file mode 100644 index 0000000000000000000000000000000000000000..59d2af85a2ff47784f6f2264c803fe6cf2a8c056 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_052.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:008fc9aa23def27b2103e02e7b1fb8075efb9b1ba6a10d7296694c920dc58173 +size 56201 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_053.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_053.png new file mode 100644 index 0000000000000000000000000000000000000000..a870274f70e9adc3f61f2a6d944b1e9e991e81ae --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_053.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9fe74f2dbdcf7bba3d5c4b53fb887e99ca455fec32f71b94b84bc054287b1c50 +size 56366 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_054.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_054.png new file mode 100644 index 0000000000000000000000000000000000000000..679d734038c65452c40ddfe4ecacb91bf0ee4fe6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_054.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d39166b9376f02b573f43b09915fabfcf8918794ca764882ae41da269ce2ed8f +size 57357 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_055.png new file mode 100644 index 0000000000000000000000000000000000000000..0ad0583f867ba7ed5856bc07905fa75f2e46e076 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58d91e2f1a2bb94d27cfdf49d0559d2af50327b9d17a6cebe1a6e556b4c63921 +size 58157 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_056.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_056.png new file mode 100644 index 0000000000000000000000000000000000000000..629d0304d52e834ccc71e0f7e180d381d040c515 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_056.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bfa41a494218037bf218d875f408aea1fd091e76299de12945ebc24f61c764d9 +size 59129 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_057.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_057.png new file mode 100644 index 0000000000000000000000000000000000000000..bc44231dd10959b3bb406e735dd7b3ce553093c0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_057.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ed106dbffa34e8ee8fd75008dd61187f62ffed2e9d65739922c91f5e0aacf03 +size 59743 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_058.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_058.png new file mode 100644 index 0000000000000000000000000000000000000000..ed04ee6e8ec9af64d909a9e387a4617491c19f5e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_058.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:298682aaf8eb773c2e6834b39655577882de0ff0089f3eadf7aa14566c266193 +size 60269 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_061.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_061.png new file mode 100644 index 0000000000000000000000000000000000000000..ea394af9bf43172843a082e0b284eab3719d41f0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_061.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:209b672e1954588ba27fdbe9e3cf7fffaf61f9a8b6d70b84fb87ee5cdca1d406 +size 61163 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_063.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_063.png new file mode 100644 index 0000000000000000000000000000000000000000..e6242f6f7486a71ba31aab0d96d896e699a56c83 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_063.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8433a6af685bbf4251cc6b26f2444ef8a24b672ab0449560e9289ab4a03fac84 +size 61953 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_064.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_064.png new file mode 100644 index 0000000000000000000000000000000000000000..d999ebd9fca73acdccf4f37c3751c6ae8a61471d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_064.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eaa05f78fa28a4383643eda3945464584e6287a5fb15322d620aae9b1f7b843f +size 62269 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..0d4b806a6e2dc9470559f4652d640289e9060f88 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:46d28c4806264f44f215c214b8411889ce7c5ef3810ce3a98907a0a8271ce0ae +size 62466 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_066.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_066.png new file mode 100644 index 0000000000000000000000000000000000000000..c24e0665a9ca20f7a068f50f701f82e67d995282 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_066.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:509b4303045a8601add19f61071269e18c4f8cb7af25dfa1d8559d84f344c4c2 +size 63557 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_067.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_067.png new file mode 100644 index 0000000000000000000000000000000000000000..704f0f772472e6dec5351ecfde074db963dbcd84 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_067.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c85191d71b897d81ba3c78dc31273dec65fb48fba058f0c3e98bd17069409aa3 +size 65245 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_068.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_068.png new file mode 100644 index 0000000000000000000000000000000000000000..f70507c94626d08adf6d71012f07040b3d1b498b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_068.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de50ac85e226684544c5138afff4f920faf5a43e311bf5bef3a8df1bd4661aad +size 66198 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_069.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_069.png new file mode 100644 index 0000000000000000000000000000000000000000..624ee7ff7f44c572af840d715f0e17d455e7f70a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_069.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e19ba017d18e5d1f288197f0ea957aabb5d7b7d8b8f0695d4602e1758ac2048 +size 66697 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_071.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_071.png new file mode 100644 index 0000000000000000000000000000000000000000..d7a42b9d959c243845ce38bda4848b8928e0f4eb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_071.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1d4de02984d687b58f9b937b3840270d8d52e17f483fc8b3c8829536ce9c9f1 +size 69061 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_072.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_072.png new file mode 100644 index 0000000000000000000000000000000000000000..4e47a5de9f4dfb6113e6686bf67d51ae60cbf82d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_072.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9ec32d7cf5754b935a0fefbb989ee9ef68563f9f5eb1b1012d62dc7e513396a7 +size 70323 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_073.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_073.png new file mode 100644 index 0000000000000000000000000000000000000000..92be1d00f97e1caa86281fc30675d94647166cb8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_073.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98d10d46d71c77490fb51e91e369cdbbe16046200102b53da9204147677ae96b +size 70211 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_074.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_074.png new file mode 100644 index 0000000000000000000000000000000000000000..d9f1aa668017fd3ae12447a128905d8377df31a8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_074.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9d66fee46fb5b459e42c2201349945090bd802f9815921948417d166595ad22f +size 70692 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..d612e02154cc2232259063dabc5099c4f1f7f334 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:770721a322f6b7c3c9fe5130f2cd8e6033f4480d1f433ea5c2964d8cf4d4260a +size 70147 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_076.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_076.png new file mode 100644 index 0000000000000000000000000000000000000000..3f77eeef7c1d8e2250a36c116901eab79ae9cd8f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_076.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ecb01562536812d075296ba67f98f37631cb1e96f4cd6ffd586ec6fce62eaa9e +size 70121 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_077.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_077.png new file mode 100644 index 0000000000000000000000000000000000000000..7aec57eb4228766b91dd367718143833385a84da --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_077.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e068e3700fc74cce7d2c0cced72b589fa17cc1fdd4b8c87b08f59f2f54d431f7 +size 69442 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_078.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_078.png new file mode 100644 index 0000000000000000000000000000000000000000..42b6287abb73bd057ee18dac15e9bd6984bde345 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_078.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f299a65acb6a3e62d628882b674491a7585c9a319913f731483df423ab41bcfb +size 69101 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..a98fea4fd6f7585a6f489ffa9366918cda02185e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9fbefa3bb5dc60d786fbbe360038bfe6b86bd56e8116ebd0fcf5f51cfbdd3fc +size 67062 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_081.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_081.png new file mode 100644 index 0000000000000000000000000000000000000000..1594b69b76889424943ea4bc4700dc61b8ff2343 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_081.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f8f0d74e87cca89aad18371bf9cd08650ab679d75ab23b026ea2dfcc4851f87 +size 67078 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_082.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_082.png new file mode 100644 index 0000000000000000000000000000000000000000..92ba15658f42210c2b1dfa4f7a5bb7a4db605ec3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_082.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:636503bc78424abc66ae9c4f5756083d8fa59417adaf37e4c43dce8dbc23c948 +size 67048 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_083.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_083.png new file mode 100644 index 0000000000000000000000000000000000000000..708fa59c7751c004e011a6722760171fd858322e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_083.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38cd9ac111629fca31a5df43f10e086f6561d897c906670859d3c1d427a710da +size 66154 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..c7ae3303815b2f854688b90cdedf4fb32ee496a7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:729c94aec02501689a73b36572c21b101b1cb89a9cdd28fbd4eb2220eb9036a7 +size 65928 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_086.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_086.png new file mode 100644 index 0000000000000000000000000000000000000000..c606ec9ac3d85c1970929a79be23bbce292401c0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_086.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a5e295d917ae2df456702ca0e8f119eab03bdfaa486833c4efde0857f9e49ac +size 66227 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_088.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_088.png new file mode 100644 index 0000000000000000000000000000000000000000..a836e4b3c2f8eff4ad8d24f1edc7f5b541b61b4a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_088.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6fc2fc8579d4c4ea588a9569b5847d73f2d06609bf5d94716a2893557ed81797 +size 65199 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_089.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_089.png new file mode 100644 index 0000000000000000000000000000000000000000..9d0b4ff5d331a1bf53112dc08f2690da47af1347 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_089.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1532760592a052f70058981c9824454be963f00ed7233ff6eabca61f0d1c6e22 +size 64894 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..75ecc8de7412e50e3737bf7de73d5b406b676fea --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac77de039b3a66d471155f6d96c216f907e2e9694f41224d0b146aba0a6fdb32 +size 63684 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_091.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_091.png new file mode 100644 index 0000000000000000000000000000000000000000..d868bfc9132801ec30adb4d77f46ec046fdc52ac --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_091.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2486aa0cf8d9129b82adc791a6a4851c758f504832f742514f2d39c37ad38767 +size 62590 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_092.png new file mode 100644 index 0000000000000000000000000000000000000000..8dc2d2b099068023985d17e717e73a72d27799c6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e9062d12a89cf2cafa34e31800e1988a68d6fc84753f0be36ff227716eb6d3d +size 61547 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_094.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_094.png new file mode 100644 index 0000000000000000000000000000000000000000..de17a43f985c61d396e9a5ba06b3f813a34116d2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_094.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b7c983c1b7d3ac8dd8ef4534b505013dcd9ad111a7e48c232dcefbd145c1574 +size 59353 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..740360eb589b5853ecaef427739fb9b10985028a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f44f3938e2217c595d526895b128d9826485307c5147f188e1e15e031a997966 +size 57804 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_096.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_096.png new file mode 100644 index 0000000000000000000000000000000000000000..12f8ab23019d42b4e5732e0e76b3cc0d5093b95d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_096.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c134dcfbcec115f58172e4ae9744dbe8882f7d7cf159f95fe00ecc5e858dd589 +size 56999 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_097.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_097.png new file mode 100644 index 0000000000000000000000000000000000000000..d950676855a2bb49b64d94c785223ca463067c86 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_097.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be9f6d390f153ce279bce4a8cdc84b5954ef2c49c5cd5e9bb6f57004706dccb6 +size 56508 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_099.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_099.png new file mode 100644 index 0000000000000000000000000000000000000000..85b176e7fed3f26a156585f223f7ce8d1997ec8a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_099.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85c456ad3b2df653ccbb7e98efa436cf068c229b5f241ecf016088492e4ba9a8 +size 55862 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..eaea0fe553468dab562a15f8ae8c525e0b1c8870 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4728c0b12288b20f491610b48d439c3599a1d322e04d807247cc7d4393decec +size 55361 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_101.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_101.png new file mode 100644 index 0000000000000000000000000000000000000000..4b9b03cb709ed3eaa81193ae12359c5c9bb379e0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_101.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66b60b7654a2ac32a15848e2fbd6632e04fea8c1302437870d31243d439ee8eb +size 54991 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_103.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_103.png new file mode 100644 index 0000000000000000000000000000000000000000..890b29b8b442ca81430023404b846ba0bafab814 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_103.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78ad21d036dbf8d87aab85fc7f8acb0b41c737a727e3546a6bd9ebe64d61c18a +size 54393 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_104.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_104.png new file mode 100644 index 0000000000000000000000000000000000000000..120025a5e2d59f9be28d4f402ddace35a37f0caf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_104.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a848cd7ef3d1ce8ece566f6f5bfce56e8b2384b4de6f4c8c6bcff8d501f657b +size 53858 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_105.png new file mode 100644 index 0000000000000000000000000000000000000000..0fb02175aa40661cbf84d6c8ad48df777a107409 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d8b0c851bc58f1e7d69764b76e88e24908ff4232d84571353ea2d1f0452879d4 +size 53007 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_106.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_106.png new file mode 100644 index 0000000000000000000000000000000000000000..040027618c5a9849a1fd53422ee3d44ec857d30d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_106.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a421dca7ca68850a619ccecff2c513d2ac4a618cc3db5fb376c6693ebb7b4deb +size 51595 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_107.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_107.png new file mode 100644 index 0000000000000000000000000000000000000000..fd37be4388740d7a09f651ee9dee5ab7d1094494 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_107.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a3a74e56381e6df88858c7aef4860156f782c764ad9cfca3d8465a327951212c +size 51534 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_108.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_108.png new file mode 100644 index 0000000000000000000000000000000000000000..6898bb1ecc84c602795e95dd1a286cdd6e99c3ba --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_108.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4af343c204c7f1b2509fcd762b51af88f1578d473672fe76018dd2fa5794a09b +size 50632 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_109.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_109.png new file mode 100644 index 0000000000000000000000000000000000000000..96fb1f203f67d0a3de395cefe773f0b06663c719 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_109.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:32b7c10af73541cbbaad4a60d87ee93265be9fd89e317afb84df0a218ca86e89 +size 50670 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_110.png new file mode 100644 index 0000000000000000000000000000000000000000..f76c80ae306d945590a384b37a003c2e9feb8cd6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca35853b4fa16f529136c6af8dc1b742028f3fa1039ec6c9094a257106e32094 +size 50535 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_111.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_111.png new file mode 100644 index 0000000000000000000000000000000000000000..533f321e45dcd5109add97d9db045c9454d58447 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_111.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28324be26d0b6d72ea89aed364938cc1f6d996ebcd32ff10eb540924acb9c802 +size 49739 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_112.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_112.png new file mode 100644 index 0000000000000000000000000000000000000000..f7415c250689790b1a6ddaec69fc5d90c7b67336 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_112.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01cec9fa134f05a3d1fd45a132b80aa63abacef542de4bf2101cec97bddc003b +size 49841 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_113.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_113.png new file mode 100644 index 0000000000000000000000000000000000000000..bcf0ec28588eaa6385a7ddb8f6aa69d257288ee7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_113.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:239cb41afc959e6f55d0356768f8acfee05df9be22407dcead165b790dc01108 +size 50384 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_114.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_114.png new file mode 100644 index 0000000000000000000000000000000000000000..6d78f67bb41817616250417da4c4632fcf065d59 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_114.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b338da1e0b904deb5a3deb8a290db45295eea6d931ed78ed1f13ac0af2e3530 +size 50760 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..8f2227459b3e7b98ddf686ba09c012b0669f2556 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9654bed56251020852281585e0438b3b7321453b062e25e736ce1c75c484d6a +size 51541 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_116.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_116.png new file mode 100644 index 0000000000000000000000000000000000000000..e898a9d34623ec3eba894bd2e5159ab8b7b2e059 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_116.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:823b0a4a16e3b6462364546bd2019b0f96d41a87a84577a5f73c98acc7fc915c +size 52386 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_117.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_117.png new file mode 100644 index 0000000000000000000000000000000000000000..88a398918c264e63cf2949536b34cdb10b2d842f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_117.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de876815c74d392472892ba0eaeeb08e49c75d7742a4bd27e878a00c6616cfbb +size 53704 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_118.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_118.png new file mode 100644 index 0000000000000000000000000000000000000000..2e9ed4ea22783908a609cf0f217b4bcec267c033 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_118.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:919b6cf4d708297cbb21d660d5d2fbd354c5b99f53a853b0c1c42a5a1b93f882 +size 54321 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_119.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_119.png new file mode 100644 index 0000000000000000000000000000000000000000..63dd0fe73753c242058a1e8dddaab90793bd813f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_119.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a6bc5af16e68904eb8fefa4d3e0631e1b63d574034082961b09d105f1afb97c4 +size 54822 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..8a83f5b3c5a9c1a4b92729140d92a0b9dfd46cd1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/mosca2/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e526fabf95e3debd744f023b728d91748f5aa187f6173237f286e0e22a486a8c +size 55003 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..84bf90ba3924d5010cdadceb98ed6a6609f52ccd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e1749094b69dd86bd53c36148215e6725d852da18a7dc85d743a3a5b5e483b8 +size 320329 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..d41b8e1a1636dbfdaa2c69b8657e2b93a7f6c79b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:604dc1a29c40f4c28a641c94c5b7aff36c96fe69cef3fb30c22e8c5fb298d019 +size 50730 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_001.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_001.png new file mode 100644 index 0000000000000000000000000000000000000000..b5c83edee11ccd1ac578655d736f0a3612c84002 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_001.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c892bf629f561a32caf329c43498381dffd749c690c04a4b542c9e1ec7d6aef5 +size 49132 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_002.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_002.png new file mode 100644 index 0000000000000000000000000000000000000000..cd2fc3d27c0bf7c805e93500b23b14d3af6c0b22 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_002.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2dc1d992fd2503b1ac1c61d5bb2e2773d666ce0021a47215b9f9393faa7bf40 +size 48479 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_003.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_003.png new file mode 100644 index 0000000000000000000000000000000000000000..26add197b49c3ef3f03e358af11bc3aaa1dc195a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_003.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86759a07560139de5f3001cea0485540bd3279803eac87acc26a1800ffd5292b +size 48503 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_004.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_004.png new file mode 100644 index 0000000000000000000000000000000000000000..df7bdbcd83ba5779d048a1135b516b184d660d1e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_004.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cdd6c727a5c144b1b8e74d35480fec175bc5700c2c52fab776de9fa5e49272c6 +size 48808 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_005.png new file mode 100644 index 0000000000000000000000000000000000000000..cbb784d43f344d7ca99ad0676758bb1ed7f5b16b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c24b7699b67dbb210ecf9c07d54bb6982e741eb9b45bf74ef26e4956acbd241 +size 49277 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_006.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_006.png new file mode 100644 index 0000000000000000000000000000000000000000..53f930213592535e6ddb8f3424665b8fa68731d7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_006.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a2e34b4a89ef14182d6a11a5c6876681db7d884fcb27f43b9e27b6f9908577f +size 49872 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_008.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_008.png new file mode 100644 index 0000000000000000000000000000000000000000..4503bc29ec3ea4614fe4528c05085a6c6135a5f7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_008.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:309f1cffe7ab436cf4cd63555b0da422f4750fb9ce63e74a6a2990c2f204fa12 +size 52956 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_009.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_009.png new file mode 100644 index 0000000000000000000000000000000000000000..9e040f0097df1d5e4b1f489e61df6c5d106d70e5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_009.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2517d4cd6b941cc8b6dde62951bd623f6e650745108665f4cc4f2bfcea98a51b +size 54282 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_011.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_011.png new file mode 100644 index 0000000000000000000000000000000000000000..6645941248151f88cae3c533efe8156c722f5c35 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_011.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba979898f31d2fb49a6dae87a01816dc352cb122794b7bc63a16843d2e28d78a +size 56843 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_012.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_012.png new file mode 100644 index 0000000000000000000000000000000000000000..6b1aad261e5346ba6d8677526b87d42ec4a5f032 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_012.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f629065fee83671bbebe933d86d12fb5a928f207c8f30f6a87d6131b69f417ef +size 57303 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_013.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_013.png new file mode 100644 index 0000000000000000000000000000000000000000..0a22222e36b104c765f398887fae439380d4a53f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_013.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67ebbc0d347db566752d9ceb7df7cec7c3294e1e7bcec8d8e9b1409066d2dbc8 +size 57557 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_014.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_014.png new file mode 100644 index 0000000000000000000000000000000000000000..1bb68786d1da0e09405bd8c75afa23709a39cfa6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_014.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b77654f368d905d65a9b0cd974f313db5d057279a12df7990024d7bf355127f1 +size 58061 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_016.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_016.png new file mode 100644 index 0000000000000000000000000000000000000000..6281c39cc3f0e53ea07241317405ee235a6259ab --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_016.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28899839dc5cca1375fcd1d3c7f6bcea53229f25b3f9e4891eaf1d011a5a57b6 +size 58736 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_018.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_018.png new file mode 100644 index 0000000000000000000000000000000000000000..3a665a59b57ab14dfd997e928133dbb2412d48b2 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_018.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df10f3f5250c58d52b5a41e3b9f5f44ec7bad152e52866214ecdaaf3fae0ae29 +size 59772 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_021.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_021.png new file mode 100644 index 0000000000000000000000000000000000000000..4c226b47317f47a9543bf106d48716f34193400e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_021.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77db5aeb2605deaa101c357fcbd718bd2e375d6daab12e22d61d93b4a44edf19 +size 61654 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_022.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_022.png new file mode 100644 index 0000000000000000000000000000000000000000..7c6ed49fa3abd7d8a26a06aeb3098f3f8ca5a8af --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_022.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a072a79596e3134a5cf68714b4432f5b62d36617324aac1649f5d6950067c75d +size 62313 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_023.png new file mode 100644 index 0000000000000000000000000000000000000000..795380ceb1c5cacb26c788c42d5d007cf5f21851 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1761905738b0ed4fa3f58dd52babc158545c50db1c1a59b88fab25b28f335c9 +size 62031 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_024.png new file mode 100644 index 0000000000000000000000000000000000000000..d980dbeb3ce665e4cbb53afcf16282df17e76561 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4ff09ff1782c292434d3a1eb22ad3aff2f129bc8ffb42942e1c68f8226205f3 +size 62361 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_025.png new file mode 100644 index 0000000000000000000000000000000000000000..316867e58a8e9459e3bdba5552617ea29e863491 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e722b6e587f0ed50ecb182e0b253358e3064d5bdfe7c53f5c81d93fb75692767 +size 62789 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_027.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_027.png new file mode 100644 index 0000000000000000000000000000000000000000..5e9a6e78bac37202f7f378e368cab008596af6a5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_027.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:51b86cb0cdb92ea29a8c8f40277e58fcba7b5b438ce932edf92554a874a49e02 +size 62399 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_028.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_028.png new file mode 100644 index 0000000000000000000000000000000000000000..5a4afcdf401e08f22a393d77759a55a952a14988 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_028.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c499d6f05f874a558f89931a601c7fd31f5b87dd175b2c9186b37bbfd844ff04 +size 61089 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..afe12fa75c401ebf58297e0d1fc42b2557c6ce36 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e1144f516944e12e24baa7b2f66da030ffecab2c8a50b0b540a0881184a04a6 +size 59931 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_031.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_031.png new file mode 100644 index 0000000000000000000000000000000000000000..c17fa75890d3ced89e2904e9d0560b96f8be52b3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_031.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:270e283671b5b2c82f95107ad6c1569ef7ce377df1567a388111d095301e349f +size 61720 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_034.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_034.png new file mode 100644 index 0000000000000000000000000000000000000000..059ca2bd6b3ffada6ad5dbe2b0be221c9d5361fb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_034.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5140d9733a8cbd45ef925cb4c1daebfa4a7f52291812521d2ce50e76c85927ce +size 64368 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_035.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_035.png new file mode 100644 index 0000000000000000000000000000000000000000..b256746e065715135c621df91ee6dab572098981 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_035.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8f21d8e4f5569aef9081218b1e266a03be95a25294ac9528c22a94d382f37db +size 64988 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_036.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_036.png new file mode 100644 index 0000000000000000000000000000000000000000..f6f62a573b4cccaf3f1de7686e091fc59eb5c196 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_036.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf80e1abb6bb397d67d34905c4bbee7309741ce892c3515489d1e86a8a7c9b8b +size 64760 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_038.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_038.png new file mode 100644 index 0000000000000000000000000000000000000000..dfccd610e3073d306a13bde939150114a1b6e016 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_038.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7038465eb56ae58415bf804546b0271a6e3ca7065277ed7001ee9c9c6db08167 +size 64014 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_039.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_039.png new file mode 100644 index 0000000000000000000000000000000000000000..7538c8f79978f3ea9c454470e683f330d4982404 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_039.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8a7dfdcbf67102ba513770d153d25d402c349774472f211b973ef74c80f5bbc +size 63508 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..7731e1907d78ac82bee3a78d3ec18a5cc8d4e1b9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27d02adee046a671f7043afaa9bc2e3511cd301bba1431eb721b6637ad2144dc +size 63047 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_041.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_041.png new file mode 100644 index 0000000000000000000000000000000000000000..e0c8ea141b7ca5ce04d867341c1e4b7b645c1cad --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_041.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57bc3c5621f287a9530b870ba8a219569e279777778d9d63bc0bfb6dfb40c55c +size 62629 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_042.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_042.png new file mode 100644 index 0000000000000000000000000000000000000000..707469658c99f7729426dd7cfece6d6ff3d2508c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_042.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33d9a74bdb5b90015e3cfdd4ff0bae9c8ecb46b6d091d433a79980f44569649a +size 62045 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_043.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_043.png new file mode 100644 index 0000000000000000000000000000000000000000..7669fdbb43bc937fe28cd903af78e3d2f9e287b8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_043.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aea56088f8e8be1320a0f8ba47a07a5100e64792135bbc1a652ef19c70d95cee +size 61308 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_045.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_045.png new file mode 100644 index 0000000000000000000000000000000000000000..e9869d204895a257c82b9faceda6cbbb8523a889 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_045.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:825730c38b080cfd9f00c100f8cf1161683668253f7742bb475551a9c96e8abd +size 61215 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_046.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_046.png new file mode 100644 index 0000000000000000000000000000000000000000..2ff6c7a1c4b2da24f1d2334c0b32607babdaf922 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_046.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e83a9b525b8bdd44ab405c5819cffc6bec5665eff0f76b7ed55632375fe5e0de +size 60823 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_047.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_047.png new file mode 100644 index 0000000000000000000000000000000000000000..27a493a55c3f811f4cb98b93cb66525c62497ca3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_047.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f93f982c2423272598cc6c8cca1b2fd95f873969c6e69d117a7a815c15c29b90 +size 60280 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_048.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_048.png new file mode 100644 index 0000000000000000000000000000000000000000..bee82c97092b06c059b42d69711e020de5abd9b9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_048.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:495c5355113d68fe2911f15018bca9f214236a38d602217ae370d091965929a6 +size 59231 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_049.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_049.png new file mode 100644 index 0000000000000000000000000000000000000000..209b6b4e3152ce5f3be0291fbc8915e3f8bdd774 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_049.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:991ec38973fa0ddb95b8adb63c8a93c4efce0e5a92009b4a53b080c9bd7be6b6 +size 58784 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..a32f69827a60b557496734127dd137141e8be437 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c4a38a479ba0823e5a3cea2b9e44166651ead10e0aaede407ba919b0a19f218 +size 58521 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_051.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_051.png new file mode 100644 index 0000000000000000000000000000000000000000..56616cfa4926576a3fd5c66d0a69b2c767c32395 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_051.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:949ed9b9a6d6ac9900cee0ca67371f2513c2b88a1449cfa8b88a8d5ae0f097c8 +size 57096 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_052.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_052.png new file mode 100644 index 0000000000000000000000000000000000000000..dbdb7c856b09a0a6e100700011e041e160dd5b6a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_052.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01b484c4f475adf7e0046ee6bdbd87a0a6653cd22ee67500d99d2da4afa1fe81 +size 56221 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_053.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_053.png new file mode 100644 index 0000000000000000000000000000000000000000..3bd55c7b2d06587184fa4f2e18ff43fb8a682cd8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_053.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8072a4e4c4fcf6ec45a0562f3e28da8c847f5d8900142cd6a6fd3de17381ce12 +size 55767 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_056.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_056.png new file mode 100644 index 0000000000000000000000000000000000000000..401b915b011c8d774a9e173a550b31b90eecbda6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_056.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:382d47014ee1d536529d96bceb310aaf0907dad98f1825aebd9081658833828a +size 57910 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_057.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_057.png new file mode 100644 index 0000000000000000000000000000000000000000..d2eaa70f8f04b3257e00baa2e339a8c5960557f5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_057.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7aef7f0c0db153abf13158f11956e8324ac7abf7e1633e4ee4c9039e836098e +size 57933 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_058.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_058.png new file mode 100644 index 0000000000000000000000000000000000000000..c22fca1e032f6c6ff79df190e20d3f2531dbe525 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_058.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4e51a7c9352a3d401d50e70d3909f263a48387e854e349eec4a1779b7b3a4ef +size 58731 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..5fe0f97246b4e1cb3a088b8bdd448fa1dee4f5a7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6cf45fc2d23f3dd4e784fbfab5c4862382d687cf75e3ab7615527dfc0b138fb9 +size 59309 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_064.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_064.png new file mode 100644 index 0000000000000000000000000000000000000000..5916761bee1297a14cf8bd7a98d59dd85036f2c7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_064.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:62b299f824ccee5ae062853989ad86f10d096281529cb379fe236cd80d661f22 +size 62028 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_065.png new file mode 100644 index 0000000000000000000000000000000000000000..313f926d048f785e88c48645f09427124e44de59 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59dce056ec1be4bfcb388cff358e091263581929f1f83d007ab9dc1bc10a36c5 +size 62494 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_066.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_066.png new file mode 100644 index 0000000000000000000000000000000000000000..ad41f0b135856b5b33a82ffa888a531e020433a0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_066.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a4f7d20b7912b08ca2f26a07601698fc7d6e1ca4fd85f67648ef0f79f1afeb4 +size 63295 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_067.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_067.png new file mode 100644 index 0000000000000000000000000000000000000000..93a9278cfd28fb137de2a09581c5d25ba6b8ed0a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_067.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78957e4107d4e2c6f41bc19f724b6264ea28b189ecb10ea134eaa9c6e6c7afb2 +size 63244 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_068.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_068.png new file mode 100644 index 0000000000000000000000000000000000000000..7da76378d7684335b58eb6214005139beec162b5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_068.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:193e8f2be99b7103cd055145054838abce2925401c26abca490cb8ba2091d2cb +size 63754 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_069.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_069.png new file mode 100644 index 0000000000000000000000000000000000000000..532e8b1ff13f0ffdcd8982ff322aa7a72d28bdc4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_069.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:245d2bbd83a9de4338d247d8e1a4ecf5543dbfba6eaa2dbe57c98d7168896081 +size 64428 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_070.png new file mode 100644 index 0000000000000000000000000000000000000000..9268f013c6ff0fcd24308a7d2351f228bb1169d5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:583019e10294c3abce77da78652626b082e8829f529239b827cc23808157b948 +size 65570 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_071.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_071.png new file mode 100644 index 0000000000000000000000000000000000000000..ab48b493e47a71a6b82c14b7a10bac964ed84276 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_071.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c6987d1d6e2a1d67ecdd44c468e9baec68957175e1be0736aa18cf6e8c670ef +size 67080 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_072.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_072.png new file mode 100644 index 0000000000000000000000000000000000000000..1533d40147f20686cdad1591752b073067f98b94 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_072.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e4409eb536cb04f675221bb07de69a9268260af504305dea482cba35a77c718 +size 68552 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_073.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_073.png new file mode 100644 index 0000000000000000000000000000000000000000..efd9a49c66547942353a3bfccaca723e5f59222d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_073.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:828b6b19c171780c5eb86e49eeefc5a9eea62aa504d786eda0ff19d6d3b1274c +size 68783 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_074.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_074.png new file mode 100644 index 0000000000000000000000000000000000000000..f2094aff380e4e43d7a735280c1f759c63e06292 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_074.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6d0e2ddb8c79a7fe71c3df3968f008ad4de7f74c91a248dfdb0e7c0843d1abe +size 68647 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..3a2cea92af0c8825d45b1e9245d5e1018a5bc067 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8884bd119dfaecbcffaf484504326bfd581e30af9c3cefcb12104f0d76abaae2 +size 68607 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_076.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_076.png new file mode 100644 index 0000000000000000000000000000000000000000..bca9d5b0025871eceffce2b9754d252f7fc46de0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_076.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5cbbe509cf1f5f645455631876d8c5a41347494e7ef5ddefa0a3a60ebf6435a9 +size 68107 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_077.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_077.png new file mode 100644 index 0000000000000000000000000000000000000000..a9c62f9723d8a37eb395d6f6673c07902b5db475 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_077.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fad11707f807cd1b90acb2922218c8c01048dbe99d6f96efb035764b42a021f +size 66937 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_078.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_078.png new file mode 100644 index 0000000000000000000000000000000000000000..172ca589972cbae60c1fb9407bd53f6d3399723f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_078.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e3ecef0e6e1bf9cff8a882027cf88ba3baa94cf560cdd50a782516526a81c09 +size 65925 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..039898e74a35b22a94a110add19c9f4ead0f8f2a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a346091f926a4d5880e82e06a791e1fd82cf7ffb818b84864577728f31464ba7 +size 64347 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_082.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_082.png new file mode 100644 index 0000000000000000000000000000000000000000..2c818f9c01db2c8b1b2254e5c618729d4532ff68 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_082.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:875b4d77f2303a1ba77c2a99d75c9099e2328c96d261a4605ae74599c1afc409 +size 64331 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_083.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_083.png new file mode 100644 index 0000000000000000000000000000000000000000..f10730877872467cbc968505a142a4c4a3279773 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_083.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22a0c990f71944d63da71d9a0ec64313e27f8f01f98ce4164033baecf987540c +size 65315 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_084.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_084.png new file mode 100644 index 0000000000000000000000000000000000000000..5097b034dd117881ccbc1972ab191e6c7f36ae81 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_084.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad53375a8cbbb0e1cd45cfcc7ee41e5d3fc5aca35beeda2aaa6f0fdd8bd8d2b3 +size 66415 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..b42d199eb122ba92e5e3d4f177c3f787a622c1f8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cfc4d75992c3ad4bd4bcb5961e678835dcc92a541efc5552f22c7f8f6a43c27f +size 67035 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_086.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_086.png new file mode 100644 index 0000000000000000000000000000000000000000..a3e8e6fe33f41b3807934920f6e9d05c5f36c9b8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_086.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4955553b5cea6b5ffadb54233ca84e901d901c07d205e81178f401104350958 +size 67776 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_087.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_087.png new file mode 100644 index 0000000000000000000000000000000000000000..694b221bab52f56fe1045a6d58f8af9b7693e073 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_087.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:715d005c078175e350b8a156077ef7cb447f6cbccfeb4955d2631a847a7154a8 +size 68775 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_088.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_088.png new file mode 100644 index 0000000000000000000000000000000000000000..3506e11d951cfd51c4414123e66874b6687f79e7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_088.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:baba450942e83738b0695daf39588ae667caa00851fd226f032be6e6122b1285 +size 68462 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..9dd9b7104b306c73e6439dca29161bb5eddb6312 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ab2cacd2a04be640502d01ae6b5c82cf5efa05de67a852d735f39e6a2a003cb +size 67914 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_091.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_091.png new file mode 100644 index 0000000000000000000000000000000000000000..868e1dbeab84cfb140146f83c91796a824ecb531 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_091.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ed18d30108cbd70da76a588808258b32926d0dc5f1bfc879f3884bd0c37627d +size 67391 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_092.png new file mode 100644 index 0000000000000000000000000000000000000000..d8ebcd30fd0a0d01aab7316dd71f5e21eb656c08 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b796343e232f376af2a8bba7f428126556d5d0c1f5ac84b25ce5c3729b52ac9 +size 67132 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_093.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_093.png new file mode 100644 index 0000000000000000000000000000000000000000..1b3592880b32d8595f5ed661262148ce5a24799b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_093.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1c3b3ef07f73abfe8467205ab1b70a78563cd35eccba389e1ddbc58660bb891 +size 66792 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_094.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_094.png new file mode 100644 index 0000000000000000000000000000000000000000..08311595b3622b099c20dfab8a9885a08cc4fd87 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_094.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43f5f8b583a58e5327b9228f88bb593a68c53f6463662fee182fe94beaa423fe +size 66521 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..cc1796f2f0807f914e69aec1e1873beec440f749 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2d50fa85b30d59f08484cd5668313c89e30e1434022fb95385a12b53ae3a3b6 +size 66128 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_096.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_096.png new file mode 100644 index 0000000000000000000000000000000000000000..2634a6a21523585c237c138994d173a42e4f60cb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_096.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71cfe6b9a251ed42c9473d1c0bc8d83cc759f2ea60854608701d3fc685425696 +size 66130 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_097.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_097.png new file mode 100644 index 0000000000000000000000000000000000000000..be59a810958c19062c456a885d05a226a1b7df08 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_097.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d44b8f5eeb3f43d894c142d49f546814b41561afeda246677ffec9bd5bccba9d +size 65071 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_098.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_098.png new file mode 100644 index 0000000000000000000000000000000000000000..ff5395cdf804a74adbe8e4ae7a2e5cd662bfc36f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_098.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89cd0f60a42cca72710e92653c35b2bf6184e17ca0ca034277f730e34466b940 +size 64062 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..0b4d4d0dd1c7f25c0b2b10c962075a06c8895f47 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7f79e6438ed1d6a3cabb12a96bb901c02a38df42c620e1109acc81c3a4edaee +size 63550 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_101.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_101.png new file mode 100644 index 0000000000000000000000000000000000000000..bc957a7beec47e918edf772621a8eb800e19b20d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_101.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3bba69b467687fdc6c9b5c461e9708d9e41c174b01386b65a1adc82bf6ccb2f +size 62973 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_102.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_102.png new file mode 100644 index 0000000000000000000000000000000000000000..39aebb42eecf60c4b3b0d58b9c2a9bd9c72e424a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_102.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:023c0d2edbfb87061d0ea7e8d0314a0aa5b1e734eea1121179159fe2675172ec +size 62350 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_103.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_103.png new file mode 100644 index 0000000000000000000000000000000000000000..28b16945b66b92050726d41cd9736ad9d86a30a6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_103.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d009ee083092c2f8b5a4f4420586a9481b6d3c5aeab3abbe03c4d2c82e4cbb1d +size 61342 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_104.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_104.png new file mode 100644 index 0000000000000000000000000000000000000000..773ff97c65e618f7bf220b734548ab0c59bdf8d4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_104.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69530ea558bd3c3a3f498040a8ac2f82f0ec8221726cdcdf8519a93566c249ea +size 60075 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_106.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_106.png new file mode 100644 index 0000000000000000000000000000000000000000..72c34e61dc04a129582c304b60f1a21039999fb4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_106.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b68ea8c8fcfb6cbe334f6e20fe826a4ba0ab07631a0f5d9a256915a9b8f45df +size 57648 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_108.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_108.png new file mode 100644 index 0000000000000000000000000000000000000000..c8a5a5eea0a30a84d600f623d27289b2e8aa839c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_108.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f0668bd4c836bd77137433c20d230e052e7469c8e34e2b80d81ed80045cec1b +size 57118 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_109.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_109.png new file mode 100644 index 0000000000000000000000000000000000000000..77981913adfc0e179fed72d7363b69d1aa67edf4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_109.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cda3d446516ed0c6e86e7b723e5a66a98fdcdb71b0f57fc45e5c29eb18ca1017 +size 56733 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_111.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_111.png new file mode 100644 index 0000000000000000000000000000000000000000..435d5389d9f4fec21ac7f82dc799dbe54436f5f0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_111.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f1f59b477c965af38dd347d99d8288694972332c8002f824855d110cf014ac2 +size 54932 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_112.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_112.png new file mode 100644 index 0000000000000000000000000000000000000000..9fb40626e5fb05f1876bc97c97a1ebd177fbc6d9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_112.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:538c85b581a2d6202295e1957741ce34914d60ab1edcf2250b6973ebae4a571a +size 54405 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_113.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_113.png new file mode 100644 index 0000000000000000000000000000000000000000..42f67a3a56f06e47bd9779354fd204c62ffc3a20 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_113.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b1da8b25ae46e6a6d62f108f3ffb563b4bc4ae6df2af967951e379a3ed59ce8c +size 54709 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_114.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_114.png new file mode 100644 index 0000000000000000000000000000000000000000..ac20a3c056c16e22964fed5a08da20bc27a06594 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_114.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:287bfbf2e70eebfcf50bb912aea744e858f46378b68da305c30c1eb94c7a0cbe +size 54906 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..245b023c4f0a5b154f714bbc5c33274e1c87299c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00dc9b8d58f54c130bf023b06d5a21435ebbaaf2e9fb41656bbd48a95cfb8039 +size 54771 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_117.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_117.png new file mode 100644 index 0000000000000000000000000000000000000000..368e0f9f5e06b6eac79839c17ebc1dfe147babc9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_117.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:06c6494bf8d400b36d9fc625eda13f23c4f544585fee966f388e3eebc66ca60a +size 53217 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_118.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_118.png new file mode 100644 index 0000000000000000000000000000000000000000..a3df08270b38d02be647283c2c7a6c368ce42502 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_118.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8799bcff72134e63b8dc4d651960ed7782997f4747027c52a71f52442e6ef2fe +size 51979 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_119.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_119.png new file mode 100644 index 0000000000000000000000000000000000000000..8580d4bd7a0943401a219e622db743dde94dd50a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_119.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:887c75cf7eecd14e62987770fa8396d7e1809e4f4436a8e70eaac93d0c7f681d +size 51532 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..d28de7b71a6338428beefec1eb375de87c845bc4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/moscaG2D/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6161ba0a15a9cc74d2d5b5fca95a00f0e48246e362a22792060a9f4a1a8f2b37 +size 51373 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/005.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/012.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/012.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/012.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/023.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/024.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/025.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/029.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/029.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/029.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/031.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/031.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/031.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/034.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/034.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/034.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/041.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/041.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/041.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/047.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/047.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/047.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/050.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/076.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/076.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/076.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/088.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/088.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/088.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/090.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/092.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/102.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/102.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/102.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/119.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/119.png new file mode 100644 index 0000000000000000000000000000000000000000..70edf8edae43425f782d490a29388114fa7ff47a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mask_square/119.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daa8fa3cf775d9fe482b05a9832b36df91d425b537a97e0b4bcba852e15a4b0 +size 1119 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_000.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_000.png new file mode 100644 index 0000000000000000000000000000000000000000..3275a64986d0c4271b98c3a0c6ab3bba143bfb9c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_000.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a633af6b16d6c68c622355944e5ee23372f500500e0bd6b29509c37ebd243e8b +size 57362 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_001.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_001.png new file mode 100644 index 0000000000000000000000000000000000000000..2b0df87ea241f12343ecbd6852135e67a0c578e5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_001.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dda1e0b3924d762d1278b97552a0f20cc0036080d6e6e6cffafd7a09fd9b33ef +size 56182 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_002.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_002.png new file mode 100644 index 0000000000000000000000000000000000000000..a23fdb700a39c538521190a7feb74ede1f1d415b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_002.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53befabbe5d671e262e5f31ef550c04ca482451beec31da8dfec4d775808e578 +size 55568 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_010.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_010.png new file mode 100644 index 0000000000000000000000000000000000000000..43fc22ec8dbde710893b1dface29b98a5eb99c57 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_010.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b121f8d47b781e911a82e4463511f3ebfba276847c60d9905e19cb2bab528f01 +size 63714 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_019.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_019.png new file mode 100644 index 0000000000000000000000000000000000000000..e72fa6f3c8894ba07dbaba5c9a97331f4cfa236a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_019.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64d216eceafa470f41ef39a778dc233a3f992b5baa9a717595540a766f38b87a +size 71894 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..3c774e28deb48cabfd3197633a2939222f2d57c1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c028a1a5bd4b3c0c49c7e6fadfbf397f52c3b1a42ff889e12bd5e1aa44da3667 +size 71974 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..7327d9e318a7448190cd235dbdc49630fb4d6013 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd8a1ccef200ce294b430d00fc2579aa629c0c0edde46c111df3c6a4a038ca41 +size 59707 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_033.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_033.png new file mode 100644 index 0000000000000000000000000000000000000000..5bb61e0f8b498379b96a443dea3dd9dc43cc39f3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_033.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d391d31d2934073ec6e34571c4a4db20d622a9ddb66171b581872eb11f7b6654 +size 59421 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..a31951723a89d0ae7146e821c3643bf45b834030 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd03b10af51f9df7ba677c3c96a14e5282928d8eed8cbeef4b255f1c912047bd +size 60994 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_042.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_042.png new file mode 100644 index 0000000000000000000000000000000000000000..7d51d4cd629268fa669f47b6438ab68258943922 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_042.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bcd2643a4e7ae84471d4e6b7d1a242cf8365780a7ddb761ac01c08cb6d004450 +size 61067 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_044.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_044.png new file mode 100644 index 0000000000000000000000000000000000000000..503e51c6aef0b4fbedd1d22475a44c7e7eb841d4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_044.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f04239045d004523ee8b9eb80c45937944898264637fc678e4ed2760d1083ff5 +size 61270 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_053.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_053.png new file mode 100644 index 0000000000000000000000000000000000000000..29db3ceedefcb14fe7ed1e2d793ca118270c8de3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_053.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7828099ee28ff89c470ab9fc3844cc6cf214ad221910276a9cd988150e93463 +size 60170 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_054.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_054.png new file mode 100644 index 0000000000000000000000000000000000000000..3d42be7f899a39d5e409364883bd3f6ae29a4e76 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_054.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b80dc7b9e6f697beecc6f3604c7232674b0ea6eae0b9d2faf6bb97d2a62d84df +size 60268 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_059.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_059.png new file mode 100644 index 0000000000000000000000000000000000000000..4346281b71af2a7ca388bccceb35cbe7ed4f3834 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_059.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a954602c8c788d861ac6ed338874b3218665cd1d72490f095b9b17948eb175e +size 63491 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_066.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_066.png new file mode 100644 index 0000000000000000000000000000000000000000..5da99602cf5e8d567be2fca6541d9a3a32e3fe83 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_066.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de7aa19dafd650807199bb9ac209d08e07698d219569679dd142558c82df687c +size 66168 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_067.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_067.png new file mode 100644 index 0000000000000000000000000000000000000000..f2db0473c101882e10d1a88183815d23c1541c99 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_067.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f57ea00b69ad051bada585f403dbc7f1604578c0c70296d10afd3d125e6404d3 +size 66340 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_071.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_071.png new file mode 100644 index 0000000000000000000000000000000000000000..5760ff64ce56241e48738351cf85c051b1e1fb99 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_071.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b14dc9ed3360849f3c6801188a6bfcaed48603a9d2f4273ab551f5bdbee16c7 +size 68599 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..9b8e0aed8ec9d0b0995853dcd029c08792e8d37e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd4f6cf690514dc5f2f3de59a3f70406db7d42699cc44b94a937f3f7217c098f +size 64153 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_081.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_081.png new file mode 100644 index 0000000000000000000000000000000000000000..2436c54424bd0d215c80df994d83b26d5b876c79 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_081.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b6a0de8d4fab0afc42838e594d4b9759b31ea0e3dec1efedc22e5a420935383 +size 63937 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..688306f344e600ec6e5ab83a314714c9243cea32 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ddfdafb24acb8d2338a3f5c9f305269705af30e58298b237ec929ce2f453d609 +size 58211 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_087.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_087.png new file mode 100644 index 0000000000000000000000000000000000000000..2a64e45f3d14fa2fee87864665e9ccb68b708947 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_087.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9d4ea36347b49a88096125a0343c3635ce355d0d256d4f304a4274bf6b62f40 +size 56505 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_096.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_096.png new file mode 100644 index 0000000000000000000000000000000000000000..dcf80218924831a63ce165f25d55ed6dccc91f63 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_096.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec2d062e9c54730c72e27e564cf627c298449bf89efae48871c14d29a0a878b1 +size 52054 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_111.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_111.png new file mode 100644 index 0000000000000000000000000000000000000000..9d7c4e873e52ddc23c04304bbceb2432d70d8633 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_111.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7277e712bc2f048878f8d231ca1e29ab6af4f5ad4f45a0c1f83fbc9cfef8a76 +size 51113 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_112.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_112.png new file mode 100644 index 0000000000000000000000000000000000000000..4fcb3c368ad6c0bdafd79a3fcbf4a2bd0e48c0f1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_112.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76983020ff9763ccba19734b106553c2d794a9af3dcf7e94bec3c40893c1c09d +size 51628 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_113.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_113.png new file mode 100644 index 0000000000000000000000000000000000000000..6d8b7337a524d0944e467c798994a36567be7d51 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/mosca2/frame_113.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00c5261e618a5619b86ad5750bf1283c13e89d4b34832f3e75193da6cc44d2c1 +size 53226 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/combine.mp4 b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/combine.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..3a1494402374f0fb91ce6d0f73cec64627fb8f1e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/combine.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8afe01f30856f7350981fac6f58616b1c22744c5143eab3812d74bfa7a0bca4e +size 320767 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_001.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_001.png new file mode 100644 index 0000000000000000000000000000000000000000..c56f48f37738557490f0acc08a2dfb5ff6b15a3a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_001.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2fbbcbc22f8d5616b450d17e5b79fb6cf6bee1a6383c55d4af2581f84649ac7 +size 50651 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_004.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_004.png new file mode 100644 index 0000000000000000000000000000000000000000..d10c5e3de091963651a810d4394dfbcab3d512d6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_004.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b646d84fdbe60c30ecb34dc683fabae80bd2ed5081d6332c92a7acce8ca2590 +size 51463 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_006.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_006.png new file mode 100644 index 0000000000000000000000000000000000000000..2ca4e23398b4dbce31c06e8ce1d4b2c55028b968 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_006.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:351bfd839046dcb91437af7cdbca8efc2873c98b14f785e244d27a70bedec293 +size 52453 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_007.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_007.png new file mode 100644 index 0000000000000000000000000000000000000000..542ef96637430fe51672a53aadf7ed05fb9ef6dc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_007.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef73522659f1cacef89abcf8be79762e466242319897ebcf5ca2ad4160aa961b +size 53261 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_008.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_008.png new file mode 100644 index 0000000000000000000000000000000000000000..a992d10d90348d80a9fe34c92a9fa8c944cc11c1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_008.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:497fc156756927496b994f9efc437f87cb5df004466374be5ef84b9dfeb5bb3f +size 54401 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_011.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_011.png new file mode 100644 index 0000000000000000000000000000000000000000..76fa51a705e6089becf8da37cfa1f03fc4cad0f7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_011.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b29f94afe14ac6c18b9d79c2d8eee46ed926410455ca0427a3bee2aec41dd3e +size 58127 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_012.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_012.png new file mode 100644 index 0000000000000000000000000000000000000000..60516ffd1821c57b38d7a833c85d9b1ae47e6787 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_012.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c054398f25078f09a32cb32e3cf4666eb6b2f04dd78699f2fe409b9d33ef0b72 +size 59542 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_013.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_013.png new file mode 100644 index 0000000000000000000000000000000000000000..d36912301b233fa52ae257b9540eb8846b4bc241 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_013.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f13d54abe2c051c1501bd63e0e359567a43d10e1c5fa62676d1a40f19add6bf +size 60617 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_014.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_014.png new file mode 100644 index 0000000000000000000000000000000000000000..ab3d7255287bf1263800f3c05bc30fb73c0201bd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_014.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95ab0b3ba7794bffca713e86b301739318a437306016223599d299c4510950c4 +size 61924 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_019.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_019.png new file mode 100644 index 0000000000000000000000000000000000000000..a6677c5bcd9c925271673a471984f8355b360616 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_019.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c42d89d7f786177fa8dac62654c55c65cbb38876f851948d7bcb80294c4b646e +size 66826 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..bd2a2b876852eb80a29f78e35f5e8f6c662a3683 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac483855f0a3e3141e8be54d7e3b238e4e4caa271bfbb3a1d89e2a823efeb17d +size 67253 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_022.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_022.png new file mode 100644 index 0000000000000000000000000000000000000000..969d8862cf5569de70235c8c38c18d6a04955029 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_022.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:479cfaeeaba4a1241d582595bb76ad3ad93ed730eaafc1f466653588ca962f3e +size 67612 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_023.png new file mode 100644 index 0000000000000000000000000000000000000000..27dc7ec7ff5dc3d440a58d44c9f821c3161925ce --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aca6f7afac2e0486baf13edfe4c386a3f4ba449894d9dfbd926e63ec6860a0a6 +size 67453 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_024.png new file mode 100644 index 0000000000000000000000000000000000000000..9ca7acfd036c309d41d69afeae2b0855a8d5e369 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71a68782f35b5ac57395863883f64ce0032545f157b4fb99e131d783e73f6fa7 +size 66442 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..1c00bde3fcb7b838d74d12cdf207dacf9882bd0d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee78a718a7f9c7ade74a758c01c41d8a487d05aecaf916e6ec08a8dcdb3fb563 +size 60429 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_033.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_033.png new file mode 100644 index 0000000000000000000000000000000000000000..efbac0f990608d5ae8f322c11a3e587fce10fefd --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_033.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b082ef3a886f677c5a999bd68fde15044abd2782fdfc22d1476336cae3fa8088 +size 60629 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_036.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_036.png new file mode 100644 index 0000000000000000000000000000000000000000..336ab557705f4bf75cceeb330cc99ca362f2b35a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_036.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:361e37d76e44057a993cd75f94d03ccf066e3b1f8532f44bd54b8d982d1b749f +size 61186 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_038.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_038.png new file mode 100644 index 0000000000000000000000000000000000000000..97879a91f8559f77a14c5c6f42eeef4178991211 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_038.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2411a06cbb271c749367bfc3636501a71742eff3f4c7b63915bd40a3a8fd7e72 +size 61786 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_039.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_039.png new file mode 100644 index 0000000000000000000000000000000000000000..1365b8cdb5c155f66a060d7814f60f05ff9605e9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_039.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c74433d034bb51eb8dd2f8c55bb32b9d7a6b40fb2ee59db512ea3c9f2d9b83a0 +size 61736 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..1250417df66252a610798ca353cab88ad0ce9650 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:693f7cdc390ff27bf615decef8e9a87238cf9b99c27979aeec4f85442f87365b +size 61979 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_042.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_042.png new file mode 100644 index 0000000000000000000000000000000000000000..d612f898a578f6b4383d3520eb026719b2bb6ac0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_042.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4942c417fe9c398b89227a4c0ed626e27426722e72b54d7712c76dd63463efda +size 61809 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_047.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_047.png new file mode 100644 index 0000000000000000000000000000000000000000..79a4cdd996728da5c9b6eb12e69ccd44ad5ea96d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_047.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76ad102f9d59f052027d733a8eea05f2e73a9f2537e35e97cd20cb68c3e84beb +size 60965 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_048.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_048.png new file mode 100644 index 0000000000000000000000000000000000000000..7d43ab250f37d2f9fa2c72009c11c5c463be41cb --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_048.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:826af5091747616b0607b4ff367c19b023fd98e79319e1e30ac2e5f43282fdf3 +size 60520 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_049.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_049.png new file mode 100644 index 0000000000000000000000000000000000000000..0871a003dcd9d0fd07a5085eb930703cb730fc1d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_049.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffd5752dc8786e407250df3fd99fd25416b6aec0803ca72e0747b98ff635c1ec +size 61507 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..d83f337d699e22637c0ee6f8f87fa7fbbc6ef7f6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:079bf995eab9590aca341c60cf1bb1d9753f4c8877650757d084585af8fa7df7 +size 62467 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_051.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_051.png new file mode 100644 index 0000000000000000000000000000000000000000..0beb02985bf41a3b7e82fa28032d745cac0ed409 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_051.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5fe8c108f7421ac4749662270d0beb842c8679dd07fa0c5654e381966793da7a +size 60991 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_053.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_053.png new file mode 100644 index 0000000000000000000000000000000000000000..0c1224ab8016d9cd7274be22e82de6399e86d93c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_053.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:588899b8d9bf47b90a8e9414066b2f62233690eb67259c626e4624b9e2a985d7 +size 60185 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_054.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_054.png new file mode 100644 index 0000000000000000000000000000000000000000..f10807419b296e494f02434d643760ac788c2e07 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_054.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fbda20ee130d0eed8e6f5246061a8b5fa4b8d5d76cd6d642af73c0180105dd10 +size 60717 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_057.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_057.png new file mode 100644 index 0000000000000000000000000000000000000000..3a1b40bd05a2414d8fe980faaab378cf42e046ce --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_057.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7563fb1015545f462b3e8d1fcc34b72f064d2293913d78720a922c4d1ce04bb +size 62856 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_059.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_059.png new file mode 100644 index 0000000000000000000000000000000000000000..65d6d11e0f590e818295a1a5ce1680ef231dc461 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_059.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d729216e3cf4eea4124d17a3b121adb52606dbc970b20cd85e049f6b1ab35ca6 +size 64065 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_064.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_064.png new file mode 100644 index 0000000000000000000000000000000000000000..d1b15292c038639feb396f181c02703567a53453 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_064.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65fea79c15f3e62140a01839a4338cc894ea3f9350e4e3d7827dc056b9b147be +size 64912 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_066.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_066.png new file mode 100644 index 0000000000000000000000000000000000000000..b277b4cc8a508e299d059a4b8cb7369f86fcad57 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_066.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1d509578679ab662dd4a5e671494268690f6655e659c5d16006d3e98123041e +size 66621 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_069.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_069.png new file mode 100644 index 0000000000000000000000000000000000000000..57000a434a2262946631c7c093f50a3f25e6e6a6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_069.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe25e0d7642905a64cfb3c85da2de3493c013385e9bb4b2495106cd694ef65db +size 68841 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_070.png new file mode 100644 index 0000000000000000000000000000000000000000..371691627d44d47a055a885e25ca0e05066f4493 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ce62ce3e575e3a19047e4709deece76aec5eb99828ee7e5a15795d2d6c6ae17 +size 69846 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_072.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_072.png new file mode 100644 index 0000000000000000000000000000000000000000..3de915b6f9dd9fc279e7f8129e5a1b83b0ff7f5c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_072.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:099e398d841c764f706066c38598acbfe6cf04071767850b1d2333bda0f0a625 +size 71837 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_074.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_074.png new file mode 100644 index 0000000000000000000000000000000000000000..f6cd6cf503ebb20765366352ed178bea8cdad99a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_074.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da9d7c48be5edd1a02b850ef7094c89ce0ed296fc024eaee9d0cc5e486aecc78 +size 71398 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_075.png new file mode 100644 index 0000000000000000000000000000000000000000..5b7e70d48b4b2e827cf7bab73d7f23b48fef7586 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90a5610684b0d5bad1319586feedcff8b259b103a84e169b626b1a986660244b +size 70737 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_077.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_077.png new file mode 100644 index 0000000000000000000000000000000000000000..b4c1a17c1e080c7d178a1d04f1089f299bd33062 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_077.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e6f3d326adbfb012b80cbfb2e2bedbff9a5cdb4018d6f3c22c70db5ab2ccda58 +size 67851 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..19e33f6be61a591130799179b1d78206c33c7050 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7368e61c2d1fec1c956432579aeb1a779d56e4157f574be29fdefa5da8882b37 +size 65085 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_081.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_081.png new file mode 100644 index 0000000000000000000000000000000000000000..a7dbe10f5d8381907b19a76587f1e8a5e23d0722 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_081.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee1a1136e1a0002dc346b641ca82f43f7d923764f6b887c623af446420aadaff +size 64689 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_082.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_082.png new file mode 100644 index 0000000000000000000000000000000000000000..458c55d98c4bb6d62b2a911e250a7040c143d00b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_082.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a37a9d49fbc167f34e5684d2f190fface69fea93b28a440445a8458cfd9aabb +size 64584 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_083.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_083.png new file mode 100644 index 0000000000000000000000000000000000000000..fcdcec17d429550b677798720829359c30b5ec47 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_083.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27440c0d08312f200b0d550defbe98682372697018a472208b54bcb37c589c8d +size 64942 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..6252bbe70079c1976bbc4fd054d44a7c7d8802d0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70729767e453b5970eeda731bfc25818ea86d59d9ffdf5cb2f2d55280eed93b8 +size 64655 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_088.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_088.png new file mode 100644 index 0000000000000000000000000000000000000000..e6d0035de857223f1f41c1b3f2b97a373ae125a1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_088.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1cc1e606489f925f51f1e9eaddaf9471d01bdd1862c0d56a9bf47947934b0b03 +size 64868 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_089.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_089.png new file mode 100644 index 0000000000000000000000000000000000000000..2c146c25d25620cb424cd590a6567bc27cea83d1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_089.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b70815c0f8c0e305e55cc78252672280dbc1e47c00b30dbfa42253f785cdcd6 +size 65194 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_090.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_090.png new file mode 100644 index 0000000000000000000000000000000000000000..213307b948cdb7a4e51e5697192764aea454ec7e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_090.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3667c33af1b83ad62df9b8f3df728b690a4dcec32a746ee383a49f65e64436d4 +size 64616 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_091.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_091.png new file mode 100644 index 0000000000000000000000000000000000000000..7617d720f215c2c39cf7c1040952d3a353119697 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_091.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f33fd0fd46f76fa9d6aa4ed3de97d7fe0ef52b4f11f1e831f145f882410bd800 +size 63913 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_092.png new file mode 100644 index 0000000000000000000000000000000000000000..d5228f7d900d0d567f77c8ad8594cc13178ceab1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75232a811a3eae2a2dd1cb9ef55141c381bcdf8ce01d1a61e0a32c12a7115726 +size 63616 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_094.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_094.png new file mode 100644 index 0000000000000000000000000000000000000000..0f0523d60b07028846ee31edc6226dd9405cb256 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_094.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ebdd93ac2edfed81232d0cd78fa2f4ccae07d0fcc51897bf0cbbd0f0ee94611 +size 62566 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..62ceb170074e92ecf2ed9ad765766ddeaba4c66f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:739609227c9ed673fec90eb578ae448c831c5f825e2cb9081bbcf56e052a12cb +size 61528 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_096.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_096.png new file mode 100644 index 0000000000000000000000000000000000000000..dbe146a44637727aee3114e4f629d5b7eaea7b24 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_096.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:950ab1c9c09503670959112a85937300a869036586a99caf043790f3769a4a66 +size 60469 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_097.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_097.png new file mode 100644 index 0000000000000000000000000000000000000000..0a84ccd4d92487e09f5dae14c82a0fe196601de3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_097.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ae3115a32c7a05a0fc9ef8c2e6cb7c6cfabf574e99312aca85c4d4bae9786aa +size 59515 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_098.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_098.png new file mode 100644 index 0000000000000000000000000000000000000000..bd516a450a640c07d6c72b0765f6f95588e94620 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_098.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:951090a23bd686f34e90b7a81a8979bc6af70480b6f6df0dce8b417416e720cb +size 58413 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_100.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_100.png new file mode 100644 index 0000000000000000000000000000000000000000..57a3cd0576866aa00ce1c1b98ff4884d771c2cf8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_100.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dff7b0fb3a1c09a8180bb5c417421041d3bfb711b6546f302aa72d40e5b18a7d +size 57068 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_101.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_101.png new file mode 100644 index 0000000000000000000000000000000000000000..81297acbb4efd194bd85eee0b57fdcd75d7372ab --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_101.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1eea05515d27b9639dda3f651020df48d62fa7b3727ef46d46c4b28628c81e21 +size 55788 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_103.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_103.png new file mode 100644 index 0000000000000000000000000000000000000000..d665b1417caf5511fd2f07cc872c95bf6e72145c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_103.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cde59714f017562c146bd275d57fe0d39fb067efe4dbed0c3a822e8900664d55 +size 53549 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_106.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_106.png new file mode 100644 index 0000000000000000000000000000000000000000..39d228b2f4ae7062d7a06e2c7b37838b91e439d6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_106.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba8a2b6cc59a2531f54d81a41f4eb9212aef6b7800bf33dbed2c42df274d5254 +size 50211 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_107.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_107.png new file mode 100644 index 0000000000000000000000000000000000000000..03d4dcb0a0b9225a9fca994d9d66dbd360b7b4c9 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_107.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6044e59d10fe7c32b92e3a2e331f80c0b62d2e71cb40b6e772570b16b9927c5 +size 50505 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_108.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_108.png new file mode 100644 index 0000000000000000000000000000000000000000..1875ec15bda791e4410c4250011561ef6034ad6f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_108.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66942c2916e7b3ee55474a63c43687c784dd8ec044b7fc6e5aa7f05753a547fd +size 51367 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_109.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_109.png new file mode 100644 index 0000000000000000000000000000000000000000..5c21900316f108bd6bcb9d9199ae5a846aba0ece --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_109.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d532778a98b602db1e4c7b7522f7e923c98ca5fe38d0ef2df995e88a78a561d +size 50788 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_111.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_111.png new file mode 100644 index 0000000000000000000000000000000000000000..d6250dffbf52e6d11bda53d6957b609230469f4f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_111.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8090b9acbe6e67829e1510efed27cf69488f1b83ac25ae68e16499aabcd2ec0d +size 49936 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_112.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_112.png new file mode 100644 index 0000000000000000000000000000000000000000..a05831c83265bcda334158e00ee278adf5903c6a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_112.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d0feb6dd45a405844d81648ee041b68565abb98f7b293436fb355494d41ece67 +size 50737 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_113.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_113.png new file mode 100644 index 0000000000000000000000000000000000000000..32b50dd4e4d7ffd3f291fcaf6fc75f326e4a22ce --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_113.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3253ce609591fe529066338ca43df43316b415adaebcb166b8bb5be9d0b6dcee +size 51137 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..d7b7405817c6699f55ad99481c367da5c0bef0e3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:195766b80362f8d1726ca49f76a570faf7bc3d9d4760435174821033e92a78db +size 53036 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_118.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_118.png new file mode 100644 index 0000000000000000000000000000000000000000..8e53b6fef1c3e6aa611630263de7ad0dfa2aca6c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_118.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c3cd93b26eae7221d886c046553792860952e9957e8d66cb60caff3b24a9b5d2 +size 51145 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..fb84a3cee8386e4a2dac6513c18594626d5e1e44 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/moscaG2D/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3df65861c17134486e35a4fbfe5a7f535c703d975e9771250c26b6b7b03ec3ed +size 51530 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/001.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/001.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/001.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/002.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/002.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/002.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/005.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/005.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/005.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/006.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/006.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/006.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/008.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/008.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/008.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/012.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/012.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/012.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/018.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/018.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/018.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/022.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/022.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/022.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/023.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/023.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/023.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/024.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/025.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/025.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/025.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/026.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/026.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/026.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/032.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/032.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/032.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/034.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/034.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/034.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/040.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/041.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/041.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/041.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/042.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/042.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/042.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/046.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/046.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/046.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/047.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/047.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/047.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/048.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/048.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/048.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/049.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/049.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/049.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/050.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/056.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/056.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/056.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/064.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/064.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/064.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/065.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/065.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/065.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/070.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/075.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/075.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/075.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/076.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/076.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/076.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/077.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/077.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/077.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/086.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/086.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/086.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/088.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/088.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/088.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/092.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/102.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/102.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/102.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/103.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/103.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/103.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/105.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/105.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/105.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/106.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/106.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/106.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/110.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/110.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/110.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/118.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/118.png new file mode 100644 index 0000000000000000000000000000000000000000..fc14b19a9cd15e4d76aee34c11a560171c31b3aa --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mask_square/118.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884757ac0fa10321743c49c75c2b91e6708a6301c49f14fa59fd4893cc738ba7 +size 1123 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_001.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_001.png new file mode 100644 index 0000000000000000000000000000000000000000..ddf166837c3aa2e17ec9a236a7b19703a71a3dab --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_001.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:06f9c0b4597ffa65d1bbc5d50928e68ece76c17cf00bba9d4118c5526047e901 +size 58650 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_004.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_004.png new file mode 100644 index 0000000000000000000000000000000000000000..5824a3f5fd452869f5ac55625369330c73a44db8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_004.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85152d6a3ba5288902d793b1d6089ff86dd1d3d593c823cbe4d9b1bd7a0d8d0d +size 57994 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_006.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_006.png new file mode 100644 index 0000000000000000000000000000000000000000..85acd5062e3ccc1ed39252abf22294e6f58c42d7 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_006.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0985c6ba4171a620a9d445d38f74b368f5b06265a2384719c3928800c76eedea +size 55442 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_007.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_007.png new file mode 100644 index 0000000000000000000000000000000000000000..46994748014fa38e7aa9e9d97e9454bbe1a63adf --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_007.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd336912cdfc0c96196f4316cc17a47718dc33c277e738ed3db403efdab32b1d +size 54178 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_008.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_008.png new file mode 100644 index 0000000000000000000000000000000000000000..8bdb6945a7644e12fdd1af7c28ca0c5197bdb975 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_008.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9c0d9912750d6c1e3c730aca34f8ff36ce1c162f81d84282f8d4dd8fa431560 +size 53736 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_012.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_012.png new file mode 100644 index 0000000000000000000000000000000000000000..6832a1b7f7980245c23b65654b034570f9fe435e --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_012.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d2fb8e96e005bb0161a4c3a965c9597394dde7d83c0b031fbfbdc65be2d3fb9 +size 54215 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_013.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_013.png new file mode 100644 index 0000000000000000000000000000000000000000..5f4a698d6ed45a0f2db6df40fce37e33f9cc1cbe --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_013.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:000c7d010d25bb15c4f4d070c2f5bc4fa24a9113e51693b275b3f9c753caf459 +size 54641 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_014.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_014.png new file mode 100644 index 0000000000000000000000000000000000000000..ebd59c4baf046a08e15b82525cdd6b7ce90b3772 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_014.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fadefa2d66f8aa9b86b538c5e6e7f4970ea1075c62d2b89bead579875ca3fd31 +size 54842 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_019.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_019.png new file mode 100644 index 0000000000000000000000000000000000000000..28ae418e0f466c97a7c982e3e8d5d805f5e23851 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_019.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c00c9d9dfdd760f5571f4bebae3b0b825269000ad8dcb30818482411a10fb072 +size 56201 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_020.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_020.png new file mode 100644 index 0000000000000000000000000000000000000000..6238091972bf808728eb6d8536f703d0e150cbe6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_020.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63627526a0bcfcc6726c7057863c908e6c4e989b472924c212d1f7cc5f6fcefc +size 56492 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_021.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_021.png new file mode 100644 index 0000000000000000000000000000000000000000..5ac2222aef165308f9e4845b0f2473c84a05073b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_021.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b28e418a63646031f23b82d6d36c9513855faef416b10153ad3e210fbd32204 +size 57022 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_024.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_024.png new file mode 100644 index 0000000000000000000000000000000000000000..340423dfd15d84f9f7e2326fc23d918fae9a110a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_024.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15277810472ee2383bb966c2805dcc867ff63b9d42a2378fe71f95355bf68ab0 +size 58294 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_030.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_030.png new file mode 100644 index 0000000000000000000000000000000000000000..2156194af9e9fd14229a00c44ff177c35aecb43c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_030.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0206bfabe9e51d5819ebe2f22fd7a79152de7ec10933d9c692c5683512b614b0 +size 63960 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_031.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_031.png new file mode 100644 index 0000000000000000000000000000000000000000..ef5dd9f47f937478da1dea5d1ce6ff918d5384ee --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_031.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3855b6093434328b6690bd6ffa1c081e55c6d807cf9f167d7f77000ff740f9ee +size 64727 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_032.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_032.png new file mode 100644 index 0000000000000000000000000000000000000000..3cb5578b5edf370a3d678d4f8de324b9b102b1f4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_032.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02bb0d33b057ce1328bf43a3ec8d457e9c3dfb4294ee46aa650e665400f9b115 +size 65591 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_033.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_033.png new file mode 100644 index 0000000000000000000000000000000000000000..b6f918f7f57797d2fbe1d6d7808bf19bb7ac83c4 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_033.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:102a8d2704a46af0c041648e5773e4fdf657be6f5346d36fb996c7ec9a8931d2 +size 65250 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_036.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_036.png new file mode 100644 index 0000000000000000000000000000000000000000..9bec6ca28b13401402e988e89ebf8c61af26e4b8 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_036.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4adfe545d23367fcbfbe0a59581e3dc699ae277161026b31f5bd2086032d0b9c +size 64480 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_038.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_038.png new file mode 100644 index 0000000000000000000000000000000000000000..602983e0671d2ad39cf25e545f13300eee5a5557 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_038.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1bfa48dd674ebaebad3994adc9bea3cb67a83b84e329f5f6dbf09fb64b34bf09 +size 64237 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_039.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_039.png new file mode 100644 index 0000000000000000000000000000000000000000..56301e642dd08dc210990e161dcc949983455270 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_039.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48e72d631917f89e89fe4a6eb5390d96f827af3ab912894cefcaadb15de4ffb9 +size 64486 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_040.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_040.png new file mode 100644 index 0000000000000000000000000000000000000000..20ee8a039f08bfc27a5b9c9453e6be67ecfdc8ea --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_040.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87589aa5ec484ffa8da0496d17ec6d8008653e986f3366b46cae49dc513f8232 +size 64863 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_041.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_041.png new file mode 100644 index 0000000000000000000000000000000000000000..41a91ecc7ba5988a7ffecdda8c549b0109987677 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_041.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1012707caa8b07ce00d48ee72bb098b7848f944a22b96c8c1b8149923a699da0 +size 64742 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_042.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_042.png new file mode 100644 index 0000000000000000000000000000000000000000..75aef3e1d3c8dcac12c6df0507270f669a17b9b6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_042.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08c6ac56f47acefaf39973aaeeef3f0263cc4f746130c53f138b42108af91f85 +size 64698 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_046.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_046.png new file mode 100644 index 0000000000000000000000000000000000000000..0b9ed2285a17af7e9edf1b3c475660be051c8cdc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_046.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b57a02ccde2c76be53646e442e0125f25beb0414451054ea24c7b49c72287986 +size 65678 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_047.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_047.png new file mode 100644 index 0000000000000000000000000000000000000000..80c33ee94b22556795c4fa2606ec31fa3c40ca3d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_047.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea57a6138e2c9e1d337e23ca68935cf98c59da33360fb6904ac5483e6aa8aaed +size 66145 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_049.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_049.png new file mode 100644 index 0000000000000000000000000000000000000000..8d38e5b2f6ad15a8767c44e4f3959e775e3b232f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_049.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29a7670946eb2465b6bf647abe989f8bd6a56cb4aaf5b0a8b095d8a1b5c04e4e +size 65566 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_050.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_050.png new file mode 100644 index 0000000000000000000000000000000000000000..fa13c12a5d4b8bdf040aac45d13554155f157a4b --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_050.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6549de925c216d5a3322be7f2b9090426cc9e4f410e6b75f6e2b83c7d6c2cf5 +size 64277 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_053.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_053.png new file mode 100644 index 0000000000000000000000000000000000000000..a32db93396cc00b8e84814bd7e29c6144face398 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_053.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:993bb019b0f977a47ce7936efba9e0237cd0f69397ff867a52eea17cbd30a6d7 +size 60956 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_054.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_054.png new file mode 100644 index 0000000000000000000000000000000000000000..e84fc3cb4a4636b81a56d001815ae0653ccf49a0 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_054.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ef47f11428abee576484091408ad5975d4f4614c11c26231903fb6849973481 +size 59739 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_055.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_055.png new file mode 100644 index 0000000000000000000000000000000000000000..5698c57c8b52580658a7d6135f0c38d3cc828ea1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_055.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e07b0c7226975a3d16bbab207a41e465ce807e75dfbe9574caa58f3a767a721 +size 58703 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_056.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_056.png new file mode 100644 index 0000000000000000000000000000000000000000..d5091ee73a82bbb0ace0d380a782684d9560dd6d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_056.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:851eca18f25f5a8c0d38546cf299a626675d68bbf47d5686db26a80fa982e2ef +size 57793 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_057.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_057.png new file mode 100644 index 0000000000000000000000000000000000000000..6c258b06ebca3af811d95b9b6ac71c01e86261f1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_057.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87906d1d4ad187a160014f13fc19ceba091bc38adae589aeb6423a40b01e92e4 +size 57644 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_058.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_058.png new file mode 100644 index 0000000000000000000000000000000000000000..6d74b4f61de6147627ced4ce76c2b828b36dbcb3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_058.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:585363280583c6152e98db0f61a91ac074704dc4e9cd87d4a6016cf645077664 +size 57662 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_059.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_059.png new file mode 100644 index 0000000000000000000000000000000000000000..aaf35bf2b4abe4995a5b60c088d295e3d7385f8f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_059.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f28639e76bc9623ca0100d570063120b94fa3a9e6f0439921cd597219637e6d +size 57142 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_060.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_060.png new file mode 100644 index 0000000000000000000000000000000000000000..6df2311d7d96dbc4e727f1239b5de4968838795c --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_060.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:574534a9e8fa3053ad02358a53ce5b71af651df394d05e76144e44f2630b238f +size 56816 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_066.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_066.png new file mode 100644 index 0000000000000000000000000000000000000000..97e793e1ca34d1588538e87a886d6c850caad899 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_066.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2add25c547e35183fd5f952f337182066b697ae3a6993bb815238dfbf911400f +size 53806 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_068.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_068.png new file mode 100644 index 0000000000000000000000000000000000000000..04a6906e0fe0d2f6851de2c4eec6a81ec1e05782 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_068.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a28e6b23e858b0f64b760b953220622e355a27eed4228d5b6c8f84cf056ac77b +size 53479 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_069.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_069.png new file mode 100644 index 0000000000000000000000000000000000000000..8ddeffbe47279b179093e96d4b42f66730aed3d5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_069.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10e80dc7820f3b7d374cf16e6b533ef593402ac15814f48a0a8bf7f6aa88f5ff +size 53646 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_070.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_070.png new file mode 100644 index 0000000000000000000000000000000000000000..062658b3cbb128822b8c7acae3231686a164c814 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_070.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:808d599330f6189d62363b51f7626f996c9f5609140cba5700e8d2ba02beaaee +size 53557 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_071.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_071.png new file mode 100644 index 0000000000000000000000000000000000000000..2fd426d7773492dfc770ee1c6c1e6039e6e28002 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_071.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba88fa53f39885af65e3ee5be2bac090772f8205c449ce19ca9fa16906db9486 +size 53827 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_072.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_072.png new file mode 100644 index 0000000000000000000000000000000000000000..cbd0bb307800fb5e9c25ed7d5e8a312698745b54 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_072.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4056bd6eda8227c185b338d41744815327b040f773b29db258809f576c1180a4 +size 53620 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_077.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_077.png new file mode 100644 index 0000000000000000000000000000000000000000..c898ae5a0bec291c5eb9883d72be2b4240efe9d1 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_077.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e89b7117c13771a95533b700d92ae3ddcaab9923f1199a74bc94ebbc4145728 +size 56129 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_080.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_080.png new file mode 100644 index 0000000000000000000000000000000000000000..2382ad356651c2bfed7ce4d3cfaa72976b6ee1e5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_080.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89cbd3845f156ba625895fb2dce27049ecea86db371a8fc92518498c6afa29fd +size 60460 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_081.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_081.png new file mode 100644 index 0000000000000000000000000000000000000000..5e23a9af35fd43cd95dd8c168daf7572f343acc6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_081.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f5003b886ca86e8574aed0141f608d5898c737e72d7dac2e2e8f688688b9204 +size 61552 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_083.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_083.png new file mode 100644 index 0000000000000000000000000000000000000000..eeb97f9db4a09e331a8371adc2a580731afa5cba --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_083.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80add073d2c40ad2f21377c78de099977cfecdd2627acf4616e49e996d2f4b3a +size 59791 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_085.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_085.png new file mode 100644 index 0000000000000000000000000000000000000000..28f119a20110bc399d1fba59298bcb029f18920f --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_085.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee9a39971de53c655bd0a078629e9e08b58f644f9f4eb2083193e441282c9f18 +size 59039 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_088.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_088.png new file mode 100644 index 0000000000000000000000000000000000000000..410df9f43fb374757444b313de30d8c1183790e3 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_088.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b13c97bc551dd4fd8c2d1b145a23aeb1b2f50f9c0ca879a69d247d3f1f2415a +size 57573 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_091.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_091.png new file mode 100644 index 0000000000000000000000000000000000000000..e67841bd7aa853a7e11903199f66c9f0db0cd1ed --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_091.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7fab02b353e110b247617d7f22e52d21b242cbc1ef04973c6106a471ad5759e4 +size 59850 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_092.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_092.png new file mode 100644 index 0000000000000000000000000000000000000000..73aae5a5365274659a26df27890265198050d5cc --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_092.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:978b0961283f89b1ad253734b2162f3358ac03220e9fbdc2308cdbb264cb35e6 +size 61261 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_095.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_095.png new file mode 100644 index 0000000000000000000000000000000000000000..f4b5b9081a99eabf815e5e3977f069a0fd36a208 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_095.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4bda76dbc7ab7f8302e48919ed7f0fdb21a77ea14130e1de99f3b3b3f2c0e2f +size 63043 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_096.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_096.png new file mode 100644 index 0000000000000000000000000000000000000000..f805264a2399423b1151fdee76c3e1d40914f153 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_096.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bc8c6ee55eafaebe1848e22e8a5d643e6fcabdd8bbf6d239d971692b0cbd763 +size 62633 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_097.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_097.png new file mode 100644 index 0000000000000000000000000000000000000000..64f5b6179a2c1b5d0d81fd6253bda0a14fe0d079 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_097.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db3b74d377ea1f20f4101d7a04ba490bafba803b266f875ba93fc2036110e12d +size 62816 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_103.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_103.png new file mode 100644 index 0000000000000000000000000000000000000000..493021006c880e4658ee38b2d1c98a1b292f5944 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_103.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4975346c3c2e667cc7c3d2e89115d97d386f58cb9f11e77d0b4f23120b1f15da +size 65322 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_104.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_104.png new file mode 100644 index 0000000000000000000000000000000000000000..b28ac6b4891248dca417cf3232cf282e07c312a5 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_104.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61238bf1d2114f986b391567daa175feb7bca38bdea84d2c16472860562d4a58 +size 64896 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_107.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_107.png new file mode 100644 index 0000000000000000000000000000000000000000..c40cb073dd1bb10e1a3dfdd95da4e1cf5f7baed6 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_107.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6617539a0e416cb55898432373602e35f42bf6d1bbda35f7f6ad8e8bf2e7da51 +size 62196 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_108.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_108.png new file mode 100644 index 0000000000000000000000000000000000000000..eb33288b97be80cd883e75fa2267203ece216b4d --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_108.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36d24edcc2978ef606c06237d9c3b1a4090e4146f22042278ce3659d81e9e976 +size 61157 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_109.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_109.png new file mode 100644 index 0000000000000000000000000000000000000000..91a31053faaf85e660ee36447a659a6ce0404012 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_109.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a8fa42ef24987c87e378b84afb537ed5fd40950478ba72f08c04ba013805cf8 +size 59703 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_111.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_111.png new file mode 100644 index 0000000000000000000000000000000000000000..3df78321b4eab2f5908be1ef60a18c48f76b7200 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_111.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8214430ee00ea2ddb8d2893075c2500298048ada35579271e2ec60b6b8fc462 +size 60165 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_112.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_112.png new file mode 100644 index 0000000000000000000000000000000000000000..aad79f3c7db698aeb667c0a1ec55b31a4e0d260a --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_112.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9a1890219e34fc80e91473706a21c19b8907801dd5d8019b60d28f1018af07b +size 60791 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_113.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_113.png new file mode 100644 index 0000000000000000000000000000000000000000..e2479a054ba3ce750c463502fcf775d804cb1016 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_113.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f75aa8aee544bde2c69a2ea3c3f4e1fe1b23c812929fd5a07681835d84d9cb21 +size 61718 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_115.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_115.png new file mode 100644 index 0000000000000000000000000000000000000000..2d5cbb5530094f47ac8f02e923a80d053db5ed52 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_115.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4d48c38f61e6c8a71cd647d1a4bc6bcd94f32980195924733332cd6d3e32737 +size 63309 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_117.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_117.png new file mode 100644 index 0000000000000000000000000000000000000000..c005d09f0db50649bf973283ede0f840ff477543 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_117.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8edbc37a2fc7f0c2b3e24b8de5dd738940950cdf78266400fdc5cf25d1bb4da +size 64589 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_118.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_118.png new file mode 100644 index 0000000000000000000000000000000000000000..41a2dc4a5bf501b95f65533d3bcef39d93c0e032 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_118.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:687af191c53bc6ed83218f269a749cd28a5201eb0594e541144efb66502fabab +size 64894 diff --git a/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_120.png b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_120.png new file mode 100644 index 0000000000000000000000000000000000000000..a7f7b0ea2e8630b934b325f7103380802e363b53 --- /dev/null +++ b/000-000-02eb49fe9545406e83d8904605cead7a-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/mosca2/frame_120.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e12181d6c56faed70cc27ee1d1da2c151743523d551085cbf7f952ffd8237c4d +size 65925 diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..14cd7c94e441aa916b41486c14ba35b1d4fe9fe7 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..97491a541375514d3b5d38563b43c1c3318e3533 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..bf63b5b58c889799a7f9053007c53cca775a197e Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-041ff593cec544c19a2a37cc38c06194-000/030/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-041ff593cec544c19a2a37cc38c06194-000/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..6fded32fbe98ac55c53790aef15a689910643a64 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-041ff593cec544c19a2a37cc38c06194-000/030/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-041ff593cec544c19a2a37cc38c06194-000/060/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-041ff593cec544c19a2a37cc38c06194-000/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..3f0631842d3039c0a8f2a2ce8add88fdb1f3b098 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-000-041ff593cec544c19a2a37cc38c06194-000/060/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-014444.txt b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-014444.txt new file mode 100644 index 0000000000000000000000000000000000000000..0065c9489de2d30cb2939fe1052d1eb2a4d3484a --- /dev/null +++ b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-014444.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-000-041ff593cec544c19a2a37cc38c06194-000 --func_name main1 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-014639.txt b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-014639.txt new file mode 100644 index 0000000000000000000000000000000000000000..ef97b30cc94208a3001ad67ee2d081eb10dcf519 --- /dev/null +++ b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-014639.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-000-041ff593cec544c19a2a37cc38c06194-000 --func_name main23 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..807e3d69d50f8c8bdaabd1238442dc6d5e7743b5 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..a145eea2cc429c184fe0f7f3b31d17754b5a5caa Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..bd1e5a4ce29e622d4f434f4ca079074d16ca4314 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..d5a7a05f7affb0af51793e31ddf56ec3964e8605 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..aa17146e8acb682c3b72de763ce6be6ec1e00627 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..4c3cac3d269f4378d75ab3a3663ed18e34b9b31c Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..ccf767805d515525b3f0dc0e07db5691b4a0cc0e Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..d13d022bb6563a82654f92fe251a6956397eb7f2 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..88a4911c6bf56cf3003d42b3efe177849f8a4f9e Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..28f06b7ce65892fc2f661422fd70103d85e79e7d Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..c2412f7a6c2469c1cc6fde24c1218d29e3c13922 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..069f7abd8da72f084bb22e7f84a8056d4d271a62 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt new file mode 100644 index 0000000000000000000000000000000000000000..c4523b5a0b1b4842cc41b1be9829341bbe241bae --- /dev/null +++ b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt @@ -0,0 +1,6 @@ +Function Execution Times +======================== + +photometric_reconstruct_from_pretrained: 2109.7215 seconds + +Total execution time: 2109.7215 seconds diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2900b8b53bf8a379675ba192b354d87bd964537d Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..63dc43da49a31eb70fc40c34366e5dec6228075e Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..67a9f4a128a1b628aebf8415775ca99818694bdc Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..25d292a773cfc22582651084f0ec490769085b57 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..7877c4599d42d00df7fd0997b76b0c34f95fff29 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..37122265b512dd7f935c77cd02c13790ca336926 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..e3933d93c4db961ae51a2cab84190be38a9dbd83 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..3629f5fca1c6162e23b621b6e947ed47fa2c9af6 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..1129bd1cc4f85669a69f43769616b82f6b440007 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2b936b1861541d75da5f67d5b8be7f09999fa020 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..81662f9a7b0c7d78a0bc35f4ed49e8658773e344 Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..347675a965bafaba21e9c75dd496765b04065f6e Binary files /dev/null and b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx differ diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt new file mode 100644 index 0000000000000000000000000000000000000000..848d4e2f6c239ebb4191c0775256c1bcd37bbd92 --- /dev/null +++ b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt @@ -0,0 +1 @@ +FPS: 176.2822716484 diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py new file mode 100644 index 0000000000000000000000000000000000000000..7016b5610ec6e7552d04c0af22d9b5b124beafb4 --- /dev/null +++ b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py @@ -0,0 +1,2928 @@ +import sys +import os +sys.path.append(os.path.dirname("../lib_render")) +sys.path.append(os.path.dirname("../lib_mosca/gs_utils")) +sys.path.append(os.path.dirname("..")) + +import torch +import numpy as np +from scipy.spatial.transform import Rotation +# import matplotlib.pyplot as plt +import pickle +import random +import glob + +# import pypose as pp +from torch import nn + +from itertools import chain +from scipy.spatial import KDTree,Delaunay,distance_matrix +import open3d as o3d +from collections import defaultdict +import roma + +from mpl_toolkits.mplot3d import Axes3D + +# import hdbscan +from sklearn.cluster import DBSCAN +from tqdm import tqdm + +from lib_mosca.mosca import __DQB_warp2__, __LBS_warp2__ + +from lib_mosca.gs_utils.utils_helper import project_points, draw_projected_points, draw_projected_points_masked, valid_and_visible, overlay_mask_yellow, stack_images_side_by_side, compute_depth_gradient, associate_pts_value, draw_projected_points_colored +from lib_mosca.gs_utils.som_vis_utils import ( + apply_depth_colormap, +) + +import matplotlib.pyplot as plt +import argparse +import logging, glob, sys, os, shutil, os.path as osp + +def get_order(arr): + """ + Computes the rank (order) of each element in the array. + + Args: + arr (np.ndarray): Input 1D array. + + Returns: + np.ndarray: Array where the i-th element is the rank of the i-th element in the input array. + """ + # Get the indices that would sort the array + sorted_indices = np.argsort(arr) + + # Create an array to store the ranks + order = np.empty_like(sorted_indices) + + # Assign ranks + order[sorted_indices] = np.arange(len(arr)) + + return order + +def count_out_dict_files(folder_path): + # Use glob to find files matching the pattern + file_pattern = os.path.join(folder_path, 'out_dict_*.pkl') + files = glob.glob(file_pattern) + return len(files) + +def calculate_edge_relative_orientation(Ts): # 854 pts will use ?? GB memory!!! + nk=Ts.shape[1] + Ts1=Ts.unsqueeze(2).repeat(1,1,nk,1) + Ts2=Ts.unsqueeze(1).repeat(1,nk,1,1) + Ts12=(Ts1.Inv()*Ts2) + # Ts_rel=relativelieAlgebra_pp(Ts1,Ts2) + return Ts12 + +def get_navie_ori_distances(Oris_rel): # TODO + Oris_rel_so3=Oris_rel.Log() # (nt, nk, nk, 3) + norm_oris_rel_so3=torch.norm(Oris_rel_so3.tensor(),dim=-1) # (nt, nk, nk) + norm_ori_rel_so3=torch.mean(norm_oris_rel_so3,axis=0) # (nk, nk) + return 1.0 - torch.cos(norm_ori_rel_so3) # (nk, nk) + + +def compute_accel_loss(transls): + # transls [nt,n,3] or oris [nt,n,4] + accel = 2 * transls[1:-1] - transls[:-2] - transls[2:] # [nt-2,n,3/4] + loss = accel.norm(dim=-1).mean() # [nt-2,n,3/4] => [nt-2,n]=> [1] + return loss +def cal_rigidity_loss(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global + stats = { + "loss_local_rigidity": loss_local_rigidity, + "loss_local_rigidity_ori": loss_local_rigidity_ori, + "loss_local_global": loss_local_global, + } + + return loss, stats + +def cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges): + oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + return loss_local_rigidity_dt,loss_local_rigidity_ori_dt + +def cal_rigidity_loss_extra(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + nt=transls.shape[0] + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + # 4. compute dt rigidity loss + # delta_t=50 + # oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + # oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + # transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + # transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + # dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + # error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # # 2. compute local rigidity loss for oris + # qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + # qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + # qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + # error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + if nt<100: + delta_t=int(nt/2) + else: + delta_t=50 + loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + # loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(50,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss_local_rigidity_10,loss_local_rigidity_ori_10=cal_dt_loss(10,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global+100*loss_local_rigidity_50+100*loss_local_rigidity_10#+100*loss_local_rigidity_ori_dt + stats = { + "loss_local_rigidity": loss_local_rigidity.item(), + "loss_local_rigidity_ori": loss_local_rigidity_ori.item(), + "loss_local_global": loss_local_global.item(), + "loss_local_rigidity_50": loss_local_rigidity_50.item(), + "loss_local_rigidity_10": loss_local_rigidity_10.item(), + } + + return loss, stats + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +# optimize_overall_with_orientation version of +def optimize_overall_with_orientation_edges( + edges, + Transls, Transls_optimizable, Mask, Dists_optimizable, weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, Ori_distances_optimizable, weight_ori_distance_optimizable, + Dists_observed_mean, Dists_observed_std, Mask_observed,Nearby_mask, + W2Cs,Depths_info_perG_key, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 ###### test + + # Depths_info_perG_key [nt,nk] + + nt = Transls.shape[0] + nk = Transls.shape[1] + ne = edges.shape[0] + + # Device setup + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + Dists_optimizable = nn.Parameter(Dists_optimizable.to(device)) + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Ori_distances_optimizable = nn.Parameter(Ori_distances_optimizable.to(device)) + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Mask = Mask.to(device) + Depths_info_perG_key=Depths_info_perG_key.to(device) + + optimizer = torch.optim.Adam( + [Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable, Ori_distances_optimizable], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, threshold=1e-3, min_lr=min_lr + ) + + # Distance matrix helper function + def distance_matrix(x, y): + return torch.cdist(x, y, p=2) + + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt,3,3] + # I0=torch.eye(3,device=device) + # I0[2,2]=0.1 # 0.01 camel + # I0=I0[None,:,:] + # I1=W2Cs_R.transpose(-1, -2)@I0@W2Cs_R # [nt,3,3] + # I1=I1[:,None,:,:] + I0=torch.eye(3,device=device).expand(nt,nk,-1,-1).clone() # [nt,nk,3,3] + I0[...,-1,-1]=Depths_info_perG_key # 0.01 camel # CHECK TODO + I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nk,3,3] + # breakpoint() + + # Optimization loop + # for epoch in range(max_epochs): + for epoch in tqdm(range(max_epochs), desc="Training Epochs", unit="epoch"): + optimizer.zero_grad() + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + a1=(Transls_optimizable-Transls)[:,:,None,:] # (nt, nk, 1, 3) + a1_T=a1.transpose(-1, -2) + squared_norm=(a1@I1@a1_T).squeeze(-1,-2) + # squared_norm=(a1@a1_T).squeeze(-1,-2) # test + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # add epsilon for numerical stability + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) # full model + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) # ablation of loss + # Dist_confident_pts_error_masked=Dist_confident_pts_error[Mask] + # Dist_confident_pts_error_masked_disturbed=Dist_confident_pts_error_masked+torch.randn_like(Dist_confident_pts_error_masked)*0.01 + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed) + # mask_shuffle=generate_ratio_binary_tensor(n=Dist_confident_pts_error_masked_disturbed.shape[0], ratio_0=0.8) + # mask_shuffle=torch.ones_like(Dist_confident_pts_error_masked_disturbed,dtype=torch.bool) + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed[mask_shuffle]) + + + # Ori_confident_pts_error=torch.norm(Oris_xyzw_optimizable_normed-Oris_xyzw,dim=-1) # test + # loss_oris_confident_pts = torch.mean(Ori_confident_pts_error[Mask]) + + mask_edge=generate_ratio_binary_tensor(n=edges.shape[0], ratio_0=0.3) + edges_mask=edges[mask_edge] + weight_normed=torch.abs(weight_distance_optimizable)/torch.norm(weight_distance_optimizable,dim=-1,p=1)[:,None] # (nu,nk) + weight_edges=weight_normed[edges_mask[:,0],edges_mask[:,1]] # (ne) + dis_local_canonical_edge=Dists_optimizable[edges_mask[:,0],edges_mask[:,1]] # (ne) CHECK order + # loss_rigidity,stats_rigidity=cal_rigidity_loss(Transls_optimizable,Oris_xyzw_optimizable_normed,edges,weight_edges,dis_local_canonical_edge) + loss_rigidity,stats_rigidity=cal_rigidity_loss_extra(Transls_optimizable,Oris_xyzw_optimizable_normed,edges_mask,weight_edges,dis_local_canonical_edge) + + # Loss 3: Regularization on weight distance + loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch[:-1] - Dists_epoch[1:])) # (nt-1,nk,nk) + # Dists_epoch_edges2=Dists_epoch_edges.reshape(nt,ne) # (nt,ne,nk) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch_edges2[:-1] - Dists_epoch_edges2[1:])) # (nt-1,nk,nk) + # loss_local_smoothness = torch.mean(torch.abs(Transls_optimizable[:-1]-Transls_optimizable[1:])) # (nt-1,nk,nk) # [TODO] + loss_accel = compute_accel_loss(Transls_optimizable) + + + # Loss 7: symmetry loss for edge weights + # TODO + # loss_symmetry=torch.mean(torch.abs(weight_distance_optimizable-weight_distance_optimizable.transpose(0,1))) + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + # a_loss_oris_confident_pts=1*loss_oris_confident_pts + a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point 0.1 common value + a_loss_sparsity=1e-1/loss_sparsity + a_loss_accel=0.1*loss_accel + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # flag_train_orientation=True + # if flag_train_orientation: + # a_loss_density=5e-3/loss_sparsity # 1e-3 for pos+rotation + # else: + # a_loss_density=5e-4/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + # a_loss_oris_distance_confident_pts=1*loss_oris_distance_confident_pts + # a_loss_ori_dist=1*loss_ori_dist + # a_loss_ori_local_smoothness=0.01*loss_ori_local_smoothness + + total_loss = a_loss_dist_confident_pts + a_loss_rigidity + a_loss_accel # + a_loss_sparsity + # + a_loss_dist + a_loss_local_smoothness + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness + a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # + a_loss_oris_confident_pts + + # Backpropagation and optimization step + total_loss.backward() + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # Logging progress + rigidity1=stats_rigidity["loss_local_rigidity"] + rigidity_ori2=stats_rigidity["loss_local_rigidity_ori"] + global3=stats_rigidity["loss_local_global"] + if epoch % log_interval == 0 or epoch == max_epochs - 1 or optimizer.param_groups[0]['lr'] < min_lr * 2: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + # f"a_loss_density: {a_loss_density.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + # f"OConfDist: {a_loss_oris_confident_pts.item():.6f}, " + f"a_ridigity: {a_loss_rigidity.item():.6f}, " + # f"rigidity1:{rigidity1:.6f}, " + # f"rigidity_ori2:{rigidity_ori2:.6f}, " + # f"global3:{global3:.6f}, " + + # f"Dist: {a_loss_dist.item():.6f}, " + # f"OConfDist {a_loss_oris_distance_confident_pts.item():.6f}, " + # f"ODist {a_loss_ori_dist.item():.6f}, " + # f"Smooth: {a_loss_local_smoothness.item():.6f}, " + # f"OSmooth: {a_loss_ori_local_smoothness.item():.6f}, " + # f"Rigidity L: {loss_local_rigidity_2.item():.6f}, " + f"a_loss_accel: {a_loss_accel.item():.6f}, " + f"a_loss_sparsity: {a_loss_sparsity.item():.6f}, " + # f"Symmetry L: {a_loss_symmetry.item():.6f}, " + # f"W L: {loss_weight_distance.item():.6f}, " + # f"W L2: {loss_weight_distance_2.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + print(f"weight_normed: {(weight_normed>0.1).sum().item():.6f}") + # print(stats_rigidity) + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and epoch > 5000: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + return Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable_normed, Ori_distances_optimizable + + +def get_knn_edges(n_neighbors,Transls,Contribs,Transls2,Contribs2,contrib_threshold, flag_remove_self=False): + # choice 2: knn graph + def o3d_knn_p(pts2,pts, num_knn, flag_remove_self=False): + indices = [] + sq_dists = [] + pcd = o3d.geometry.PointCloud() + pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + pcd_tree = o3d.geometry.KDTreeFlann(pcd) + for p in pts2: + if not flag_remove_self: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + indices.append(i[0:]) + sq_dists.append(d[0:]) + else: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn + 1) + indices.append(i[1:]) + sq_dists.append(d[1:]) + return np.array(sq_dists), np.array(indices) + + nk=Transls.shape[1] + nu=Transls2.shape[1] + + i_s_knn_others,i_s_knn_is=[],[] + # n_neighbors=5 + for i in range(nu): + t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + pts2=[pt2] # just one point + sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors, flag_remove_self) + sq_dists,indices= sq_dists[0],indices[0] + i_s_knn=i_s[indices] # usage: ids[i_s_knn] + i_s_knn_others.append(i_s_knn) + i_s_knn_is.append(i*torch.ones_like(i_s_knn)) + i_s_knn_is=torch.cat(i_s_knn_is) + i_s_knn_others=torch.cat(i_s_knn_others) + edges_others=torch.stack([i_s_knn_is,i_s_knn_others],dim=-1) + # breakpoint() + # old version: has bug when knn key points are not enough + # i_s_knn_others=torch.stack(i_s_knn_others) + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # edges_others.shape # nu*5 + return edges_others + +# def get_knn_edges_naive(n_neighbors,Transls,Transls2, flag_remove_self=False): +# nk=Transls.shape[1] +# nu=Transls2.shape[1] +# nt=Transls.shape[0] +# Dist=[] +# for t in range(nt): +# dist=(Transls2[t][:,None]-Transls[t][None]).norm(dim=-1) +# Dist.append(dist) +# Dist=torch.stack(Dist,dim=0) +# Dist=Dist.mean(dim=0) + +# def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): +# """ +# Naive temporal k-NN graph construction using average distance over time. + +# Args: +# n_neighbors (int): Number of neighbors. +# Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. +# Transls2 (torch.Tensor): (T, N2, D) tensor of query points. +# flag_remove_self (bool): If True and N1 == N2, removes self-matches. + +# Returns: +# torch.Tensor: (N2 * n_neighbors, 2) edge index pairs [Transls2_index, Transls_index]. +# """ +# T, N1, D = Transls.shape +# N2 = Transls2.shape[1] + +# # Compute temporal average pairwise distances +# dist_list = [] +# for t in range(T): +# dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) +# dist_list.append(dist) +# Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + +# if flag_remove_self and N1 == N2: +# diag = torch.arange(N2, device=Dist.device) +# Dist[diag, diag] = float('inf') + +# # Get top-k nearest neighbors for each Transls2 point +# _, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + +# # Build edge index pairs +# row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) +# edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + +# return edges_others + +def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): + """ + Naive temporal k-NN graph construction using average distance over time. + + Args: + n_neighbors (int): Number of neighbors. + Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. + Transls2 (torch.Tensor): (T, N2, D) tensor of query points. + flag_remove_self (bool): If True and N1 == N2, removes self-matches. + + Returns: + edges_others (torch.Tensor): (N2 * k, 2) edge index pairs [Transls2_index, Transls_index]. + dists (torch.Tensor): (N2 * k,) mean distances corresponding to each edge. + """ + T, N1, D = Transls.shape + N2 = Transls2.shape[1] + + # Compute temporal average pairwise distances + dist_list = [] + for t in range(T): + dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) + dist_list.append(dist) + Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + + if flag_remove_self and N1 == N2: + diag = torch.arange(N2, device=Dist.device) + Dist[diag, diag] = float('inf') + + # Get top-k nearest neighbors and distances + dists, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + + # Build edge index pairs + row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) + edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + dists_edges = dists.reshape(-1) # (N2 * k,) + + return edges_others, dists_edges + + + + + +def optimize_others_with_orientation_edges4( + edges, + Transls, Transls_optimizable, Mask, #Dists_optimizable, #weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, #Ori_distances_optimizable, #weight_ori_distance_optimizable, + Transls_key, Oris_xyzw_key,Transls_key_optimized, Oris_xyzw_key_optimized, + weight_param_key_optimizable,sq_dists_key_nonkey_edges, + ts_most_confident,per_nonkeyGaussian_dweight, + W2Cs,Depths_info_perG_nonkey, + device, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 + # weight_param_key_optimizable [nk] per-keyGuassian weight + # Depths_info_perG_nonkey [nt,nu] + # nt = ts.shape[0] + + + nt_all = Transls.shape[0] + nu=Transls.shape[1] + nk=Transls_key_optimized.shape[1] + ne=edges.shape[0] + + Oris_xyzw=Oris_xyzw/Oris_xyzw.norm(dim=-1,keepdim=True) # normalize + # Oris_xyzw_optimizable=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) # normalize + Oris_xyzw_key_optimized=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) # normalize + + # Device setup + # Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + # Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Transls_optimizable=Transls_optimizable.to(device) + Oris_xyzw_optimizable=Oris_xyzw_optimizable.tensor().to(device) + + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Transls_key = Transls_key.to(device) + Oris_xyzw_key = Oris_xyzw_key.to(device) + Transls_key_optimized = Transls_key_optimized.to(device).detach() + Oris_xyzw_key_optimized = Oris_xyzw_key_optimized.to(device).detach() + weight_param_key_optimizable = nn.Parameter(weight_param_key_optimizable.to(device)) + # sq_dists_key_nonkey = sq_dists_key_nonkey.to(device) + sq_dists_key_nonkey_edges = sq_dists_key_nonkey_edges.to(device) + per_nonkeyGaussian_dweight = nn.Parameter(per_nonkeyGaussian_dweight.to(device)) # don't use for now, just keep + # per_nonkeyGaussian_dweight=None + Depths_info_perG_nonkey = Depths_info_perG_nonkey.to(device) + + # sq_dists_key_nonkey_edges=sq_dists_key_nonkey[edges[:,0],edges[:,1]] # (ne) + exp_sq_dists_key_nonkey_edges=torch.exp(-sq_dists_key_nonkey_edges) # (ne) + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt_all,3,3] + + Mask = Mask.to(device) + + # Distance matrix helper function + # def distance_matrix(x, y): + # return torch.cdist(x, y, p=2) + + @torch.no_grad() # check + def cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu): + # weight_param_key_optimizable [nk] + # exp_sq_dists_key_nonkey_edges [ne] + device=weight_param_key_optimizable.device + nk=weight_param_key_optimizable.shape[0] + + weight_param_key_optimizable_k=100*torch.abs(weight_param_key_optimizable) # [TODO]: 2000 # 100 + weight_param_edges=weight_param_key_optimizable_k[edges[:,1]] # (ne) + # assert all(weight_param_edges>0) + # assert all(sq_dists_key_nonkey_edges>0) + # weight_edges = (exp_sq_dists_key_nonkey_edges**(1/10))**weight_param_edges # TODO: to unify scale #test: 10 for block + weight_edges = exp_sq_dists_key_nonkey_edges**weight_param_edges # TODO: to unify scale #test: 20 for block + + weight_edges = torch.clamp(weight_edges, min=1e-10, max=1e10) # test + + weights_matrix=torch.zeros((nu,nk),device=device) + weights_matrix[edges[:,0],edges[:,1]]=weight_edges + sum_temp = torch.sum(weights_matrix,dim=-1,keepdim=True) + if any(sum_temp==0): + breakpoint() + if torch.isnan(weight_param_key_optimizable).any(): + breakpoint() + weights_matrix_normed=weights_matrix/sum_temp + + weight_edges_normed=weights_matrix_normed[edges[:,0],edges[:,1]] # (ne) + + return weight_edges_normed + + + + # prepare + Oris_xyzw_key_norm=Oris_xyzw_key/Oris_xyzw_key.norm(dim=-1,keepdim=True) + Oris_xyzw_key_optimized_norm=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) + n_neighbors=(edges[:,0]==0).sum() + if n_neighbors!=5: + breakpoint() + + # new + src_xyz_1=Transls[ts_most_confident,edges[::n_neighbors,0]] # (nu,3) + src_xyzw_1=Oris_xyzw[ts_most_confident,edges[::n_neighbors,0]] # (nu,4) + src_xyz_1 = nn.Parameter(src_xyz_1) + src_xyzw_1 = nn.Parameter(src_xyzw_1) + + # get sk_src_node and sk_dst_node (key nodes) + ts_most_confident_repeat=ts_most_confident.repeat(n_neighbors,1).T.reshape(-1) # (nt*5) + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + # optimizer = torch.optim.Adam( + # [weight_param_key_optimizable,src_xyz_1,src_xyzw_1,per_nonkeyGaussian_dweight], lr=lr + # ) + optimizer = torch.optim.Adam( + [weight_param_key_optimizable,src_xyz_1,src_xyzw_1], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, min_lr=min_lr + ) + ######################################################3 + + # ts_batches = torch.split(ts_all, batch_size) + # for ts_batch in ts_batches: + # Optimization loop + # for epoch in range(max_epochs): + # ts_all=torch.arange(nt_all,device=device) + # batch_size=30 + # pbar=tqdm(range(max_epochs)) + # for step in pbar: + # optimizer.zero_grad() + ts_all = torch.arange(nt_all, device=device) + batch_size = 30 + steps_per_epoch = (nt_all + batch_size - 1) // batch_size + total_steps = steps_per_epoch * max_epochs + + pbar = tqdm(range(total_steps), desc="Training") + + for step in pbar: + epoch = step // steps_per_epoch + inner_step = step % steps_per_epoch + + # Shuffle once per epoch + if inner_step == 0: + ts_shuffled = ts_all[torch.randperm(nt_all)] + + # Select current batch + start = inner_step * batch_size + ts_batch = ts_shuffled[start:start + batch_size] + pbar.set_description(f"Epoch {epoch}, Step {inner_step}/{steps_per_epoch-1}") + + optimizer.zero_grad() + + # print(f"ts_batch: {ts_batch}") + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # Mask_DQB=Mask[ts_batch] + + sk_nt_node_xyz_1=Transls_key[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_2=sk_nt_node_xyz_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_1=Oris_xyzw_key_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_1=sk_nt_node_xyzw_1[:,:,[3,0,1,2]] + sk_nt_node_quat_2=sk_nt_node_wxyz_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,4) + sk_nt_dst_node_xyz=sk_nt_node_xyz_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat=sk_nt_node_quat_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # error propagation + I0=torch.eye(3,device=device).expand(nt,nu,-1,-1).clone() # [nt,nu,3,3] + # I0[...,-1,-1]=Depths_info_perG_nonkey # 0.001 camel # CHECK TODO + # I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nu,3,3] + I1=I0 # test # [nt,nu,3,3] + I1_masked=I1[Mask_DQB] # [n_mask,3,3] + + if torch.isnan(weight_param_key_optimizable).any(): + print("weight_param_key_optimizable is nan") + breakpoint() + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + src_xyz=src_xyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask<=nt*nu,4) + + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) # [nu,n_neighbors] + # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # sk_w=weight_per_nonkeyGaussian_adjusted.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz, # [N,K,3] + sk_nt_dst_node_quat, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + # Transls_interpolation=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + # Oris_interpolation=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_interpolation_masked=mu_dst # [n_mask,3] + Oris_interpolation_masked=quat_dst_xyzw # [n_mask,4] + + + # error propagation + # a1=(Transls_interpolation[Mask]-Transls[Mask])[:,None,:] # (n_mask, 1, 3) + a1=(Transls_interpolation_masked-Transls[ts_batch][Mask_DQB])[:,None,:] # (n_mask, 1, 3) + # a1=(Transls_interpolation_masked-Transls[ts_batch])[:,None,:] # (n_mask, 1, 3) # ablation of loss + a1_T=a1.transpose(-1, -2) # (n_mask, 3, 1) + squared_norm=(a1@I1_masked@a1_T).squeeze(-1,-2) # [n_mask] + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # [n_mask] # add epsilon for numerical stability + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) # [nt,nu] + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) + # Dist_confident_pts_error=torch.norm(Transls_interpolation[Mask]-Transls[Mask],dim=-1) # [nt,nu] # last one + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Ori_confident_pts_error=torch.norm(Oris_interpolation[Mask]-Oris_xyzw[Mask],dim=-1) # [n_mask] + Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] + # Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] # ablation of loss + loss_ori_confident_pts = torch.mean(Ori_confident_pts_error) + + # Loss 3: Regularization on weight distance + # loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_velocity = torch.mean(torch.abs(Transls_interpolation[:-1]-Transls_interpolation[1:])) # (nt-1,nk,nk) # [TODO] + # loss_ori_velocity = torch.mean(torch.abs(Oris_interpolation[:-1] - Oris_interpolation[1:])) + + # loss_accel = compute_accel_loss(Transls_interpolation) + # loss_ori_accel=compute_accel_loss(Oris_interpolation) + + + # Loss 6: regularize per nonkey Gaussian delta weight + # loss_per_nonkeyGaussian=torch.mean(torch.abs(per_nonkeyGaussian_dweight)) + # loss_per_nonkeyGaussian = 1000*torch.mean(per_nonkeyGaussian_dweight ** 2) # TODO + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + a_loss_ori_confident_pts=1*loss_ori_confident_pts + # a_loss_interpolation=1*loss_interpolation # 100 + # a_loss_interpolation_ori=1*loss_interpolation_ori # 100 + # a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point + + # a_loss_velocity=0.1*loss_velocity + # a_loss_ori_velocity=0.1*loss_ori_velocity + # a_loss_accel=0.1*loss_accel + # a_loss_ori_accel=0.1*loss_ori_accel + + # a_loss_per_nonkeyGaussian=1e-2*loss_per_nonkeyGaussian + + # a_loss_sparsity=1e-3*loss_sparsity + + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + + total_loss = a_loss_dist_confident_pts + a_loss_ori_confident_pts #+ a_loss_per_nonkeyGaussian#+ \ + # a_loss_velocity + a_loss_ori_velocity + a_loss_accel + a_loss_ori_accel #+ a_loss_sparsity \ + # a_loss_dist + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness+ a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # a_loss_rigidity + # a_loss_interpolation + a_loss_interpolation_ori + \ + + # Backpropagation and optimization step + total_loss.backward() + # total_loss.backward(retain_graph=True) + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # # Logging progress + if epoch % log_interval == 0 or epoch == max_epochs - 1: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + f"a_loss_ori_confident_pts: {a_loss_ori_confident_pts.item():.6f}, " + # f"a_loss_per_nonkeyGaussian: {a_loss_per_nonkeyGaussian.item():.6f}, " + # f"a_loss_interpolation: {a_loss_interpolation.item():.6f}, " + # f"a_loss_interpolation_ori: {a_loss_interpolation_ori.item():.6f}, " + # f"a_loss_rigidity: {a_loss_rigidity.item():.6f}, " + # f"a_loss_velocity: {a_loss_velocity.item():.6f}, " + # f"a_loss_ori_velocity: {a_loss_ori_velocity.item():.6f}, " + # f"a_loss_accel: {a_loss_accel.item():.6f}, " + # f"a_loss_ori_accel: {a_loss_ori_accel.item():.6f}, " + # f"Density L: {a_loss_density.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + # # print(f"weight_normed: {(weight_normed>1e-6).sum().item():.6f}") + # # print(f"weight_param_key_optimizable:" {weight_param_key_optimizable}) + + pbar.set_description(f"L:{total_loss.item():.6f}, "+ + f"c:{a_loss_dist_confident_pts.item():.6f}, "+ + f"oc:{a_loss_ori_confident_pts.item():.6f}, "+ + # f"per:{a_loss_per_nonkeyGaussian.item():.6f}, "+ + # f"i: {a_loss_interpolation.item():.6f}, "+ + # f"io: {a_loss_interpolation_ori.item():.6f}, "+ + # f"r:{a_loss_rigidity.item():.6f}, "+ + # f"b:{a_loss_velocity.item():.6f}, "+ + # f"ov:{a_loss_ori_velocity.item():.6f}, "+ + # f"a:{a_loss_accel.item():.6f}, "+ + # f"oa:{a_loss_ori_accel.item():.6f}, "+ + f"lr:{optimizer.param_groups[0]['lr']:.2e}") + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + # # TEMP: for keeping data structure + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + # sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) + # # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # # sk_w=weight_per_nonkeyGaussian_adjusted[None,:,:].repeat(nt,1,1)[Mask_DQB] # (n_mask,n_neighbors) + + # weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + # weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + # # for i,e in enumerate(edges): + # # weight_distance_optimizable[e[0],e[1]]=weight_edges[i] + # weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + # print("End of optimize_others_with_orientation_edges4.") + # # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + # print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + # blend_dict = { + # 'sk_w': sk_w, + # 'src_xyz': src_xyz, + # 'src_quat': src_quat, + # 'sk_nt_src_node_xyz': sk_nt_src_node_xyz, + # 'sk_nt_src_node_quat': sk_nt_src_node_quat, + # 'sk_nt_dst_node_xyz': sk_nt_dst_node_xyz, + # 'sk_nt_dst_node_quat': sk_nt_dst_node_quat + # } + + + + # ********************* new trying begin ******************* + # Step 1: calculate src_xyz_1_new and src_xyzw_1_new + weight_param_key_optimizable=weight_param_key_optimizable.abs() + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # (nu,n_neighbors) + + weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + + # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + sk_w=weight_per_nonkeyGaussian # (nu,n_neighbors) + + sk_dst_node_xyz_1=Transls_key_optimized[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_dst_node_xyzw_1=Oris_xyzw_key_optimized_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_dst_node_wxyz_1=sk_dst_node_xyzw_1[:,[3,0,1,2]] + sk_dst_node_xyz_2=sk_dst_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_dst_node_quat_2=sk_dst_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + # breakpoint() + mu_dst, quat_dst=__DQB_warp2__( # [nu,3], [nu,4] + sk_w, # [N,K] + src_xyz_1, # [N,3] + sk_src_node_xyz_2, # [nu,n_neighbors,3] + sk_src_node_quat_2, # [nu,n_neighbors,4] + sk_dst_node_xyz_2, # [N,K,3] + sk_dst_node_quat_2, # [N,K,4] + dyn_o=None, + src_quat=src_wxyz_1, # [N,4] # wxyz + ) + src_xyz_1_new=mu_dst + src_xyzw_1_new=quat_dst[:,[1,2,3,0]] # [nu,4] + # breakpoint() + # for further optimization + blend_opt_dict = { + 'weight_param_key_optimizable': weight_param_key_optimizable, # (nu,n_neighbors) + 'src_xyz_1': src_xyz_1_new, # (nu,3) # CHECK + 'src_xyzw_1': src_xyzw_1_new, # (nu,4) # CHECK + 'per_nonkeyGaussian_dweight': per_nonkeyGaussian_dweight, # (nu,n_neighbors) + 'exp_sq_dists_key_nonkey_edges': exp_sq_dists_key_nonkey_edges, # (ne) + 'ts_most_confident': ts_most_confident, # (nu) + } + + # Step 2: calculate Transls_optimized and Oris_xyzw_optimized + Transls_optimized,Oris_xyzw_optimized=[],[] + ts_batches = torch.split(ts_all, batch_size) + for ts_batch in ts_batches: + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,4) + + # key points dst needed to be updated; src is the same + sk_nt_node_xyz_op_1=Transls_key_optimized[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_op_2=sk_nt_node_xyz_op_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_op_1=Oris_xyzw_key_optimized_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_op_1=sk_nt_node_xyzw_op_1[:,:,[3,0,1,2]] + sk_nt_node_quat_op_2=sk_nt_node_wxyz_op_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_dst_node_xyz_op=sk_nt_node_xyz_op_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat_op=sk_nt_node_quat_op_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # nonkey points update when src_xyz_1 and src_xyzw_1 are trainable + src_xyz=src_xyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask<=nt*nu,4) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz_op, # [N,K,3] + sk_nt_dst_node_quat_op, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + Transls_optimized_batch=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + Oris_xyzw_optimized_batch=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_optimized.append(Transls_optimized_batch) + Oris_xyzw_optimized.append(Oris_xyzw_optimized_batch) + Transls_optimized=torch.cat(Transls_optimized,dim=0) # [nt_all,nu,3] + Oris_xyzw_optimized=torch.cat(Oris_xyzw_optimized,dim=0) # [nt_all,nu,4] + + # ********************* new trying end ******************* + # breakpoint() + blend_dict={} + return Transls_optimized, Oris_xyzw_optimized, weight_param_key_optimizable,weight_distance_optimizable,blend_dict,blend_opt_dict + +def get_knn_edges_key(Transls,Mask,k): + # group by variance of relative motion + # Transls [nt,nk,3] + # std_threshold=0.01 + # max_dist_threshold=0.4 + + def nanstd(o, dim, keepdim=False): + result = torch.sqrt( + torch.nanmean( + torch.pow( torch.abs(o-torch.nanmean(o,dim=dim).unsqueeze(dim)),2), + dim=dim + ) + ) + if keepdim: + result = result.unsqueeze(dim) + return result + + nu=Transls.shape[1] + + # Dist=torch.cdist(Transls, Transls, p=2) + Dist = torch.norm(Transls[:, :, None, :] - Transls[:, None, :, :], dim=-1) # [nt, nu, nk] + Mask_dist=Mask[:, :, None] * Mask[:, None, :] # [nt, nu, nk] + Dist_masked = torch.where(Mask_dist, Dist, torch.tensor(float('nan'), device=Dist.device)) + # std_dist=Dist.std(dim=0) # [nu,nk] + std_dist=nanstd(Dist_masked,dim=0) # [nu,nk] + + mean_dist_masked=Dist_masked.nanmean(dim=0) # [nu,nk] # may include nan + # mean_dist=Dist.mean(dim=0) # [nu,nk] + Dist_masked2 = torch.where(Mask.unsqueeze(1).expand(-1, nu, -1), Dist, torch.tensor(float('nan'), device=Dist.device)) + mean_dist_masked2=Dist_masked2.nanmean(dim=0) # [nu,nk] # should not include nan # may still have nan + # mean_dist_masked2=Dist.mean(dim=0) # [nu,nk] # overwrite [TEST] + # breakpoint() + assert torch.any(mean_dist_masked2.isnan())==False + mean_dist_masked = torch.where(~torch.isnan(std_dist), mean_dist_masked, mean_dist_masked2) + # std_dist[~torch.isnan(std_dist)].mean(),mean_dist[~torch.isnan(mean_dist)].mean() + Dist_masked2 = torch.where(Mask_dist, Dist_masked, torch.tensor(float('-inf'), device=std_dist.device)) + # np.histogram(std_dist.numpy(),bins=50) + # np.histogram(std_dist[~torch.isnan(std_dist)].cpu().numpy(), bins=50) + # std_dist.min() + + max_dist = torch.max(Dist_masked2, dim=0).values # has -inf + max_dist2 = torch.where(~torch.isinf(max_dist), max_dist, torch.tensor(float('inf'), device=max_dist.device)) + # np.histogram(max_dist2.numpy()) + # np.histogram(max_dist[~torch.isinf(max_dist)].cpu().numpy(), bins=10) + + # (std_dist<0.01).sum(),(max_dist[~torch.isinf(max_dist)]<0.4).sum(),(max_dist2<0.4).sum(),torch.isnan(std_dist).sum() + + # try 1: + # connectivity_matrix=(std_dist40: + conditiona=Contribs2>0.5*np.max(Contribs2,axis=0) # 0.707 # TODO + # conditionb=Contribs2>contrib_threshold + # condition2=np.logical_or(conditiona,conditionb) + condition2=conditiona + else: + conditionb=Contribs2>contrib_threshold + condition2=conditionb + Mask2=torch.tensor(condition2,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable2=Oris_xyzw2.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edges + # 2.1 position edges + Dists_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True) + + # 2.2 orientation edges + Ori_distances_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True)*2 + + # 3. edge weight + # 3.1 position stds + weight_distance_optimizable2=torch.zeros((nu,nk),dtype=torch.float32,requires_grad=True) + + # knn graph (max contribs version) + + # def o3d_knn_p(pts2,pts, num_knn): + # indices = [] + # sq_dists = [] + # pcd = o3d.geometry.PointCloud() + # pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + # pcd_tree = o3d.geometry.KDTreeFlann(pcd) + # for p in pts2: + # [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + # indices.append(i[0:]) + # sq_dists.append(d[0:]) + # return np.array(sq_dists), np.array(indices) + + + # i_s_knn_others=[] + # sq_dists_5=[] + # n_neighbors=5 # 5 # 10 can not work + # ts_most_confident=[] + # for i in range(nu): + # t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + # eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + # pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + # i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + # pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + # pts2=[pt2] # just one point + # sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors) + # sq_dists,indices= sq_dists[0],indices[0] + + # flag_use_key_point_key_neighbors=True + # if flag_use_key_point_key_neighbors: + # ## given a key point, get the knn n_neighbors key points + # indice_nearest_key=indices[0] # nearest key point index + # pt3=pts[indice_nearest_key] # nearest key point + # pts3=[pt3] # just one key point + # sq_dists, indices = o3d_knn_p(pts3, pts, n_neighbors) # inlcude itself + # sq_dists,indices= sq_dists[0],indices[0] + # ## + # # calcuate the distance between the i-th other point and the neighboring key points + # sq_dists=np.linalg.norm(pts[indices]-pt2,axis=-1) # [n_neighbors] + + + # i_s_knn=i_s[indices] # usage: ids[i_s_knn] + # i_s_knn_others.append(i_s_knn) + # sq_dists_5.append(sq_dists) + # ts_most_confident.append(t_most_confident) + # i_s_knn_others=torch.stack(i_s_knn_others) + # sq_dists_5=np.array(sq_dists_5) + # sq_dists_5=torch.tensor(sq_dists_5,dtype=torch.float32) + # ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # # edges_others.shape # nu*5 + + # knn graph (minmax distance + contrib version) + n_neighbors=5 + edges_others,mean_dists=get_knn_edges_others(Transls,Mask,Transls2,Mask2,k=n_neighbors) + sq_dists_5=mean_dists[edges_others[:,0],edges_others[:,1]].reshape(-1,n_neighbors) # [nu,n_neighbors] + # edges_others,dists_edges=get_knn_edges_naive(n_neighbors,Transls,Transls2,flag_remove_self=False) # ablation of uknn + # sq_dists_5=dists_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + ts_most_confident=np.argmax(Contribs2[:,:],axis=0) # [nu] + ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + torch.cuda.empty_cache() + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + + # Method 3: + # use Dual Quat Blending (refer to MoSca) + # sq_dists_key_nonkey=-torch.ones([nu,nk]) + # sq_dists_key_nonkey[edges_others[:,0],edges_others[:,1]]=sq_dists_5.reshape(-1) + sq_dists_key_nonkey_edges=sq_dists_5.reshape(-1) # [ne] + weight_param_key_optimizable=1*torch.ones(nk,dtype=torch.float32,requires_grad=True) # 2000 [TODO] + per_nonkeyGaussian_dweight=torch.zeros(nu,n_neighbors,dtype=torch.float32,requires_grad=True) + + # batch training along t + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + + Transls_optimized2,Oris_xyzw_optimized2, weight_param_key_optimized2,weight_distance_optimized2,blend_dict,blend_opt_dict=\ + optimize_others_with_orientation_edges4(\ + edges_others,\ + Transls2, Transls_optimizable2.clone(), Mask2, \ + Oris_xyzw2, Oris_xyzw_optimizable2.clone(), \ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + weight_param_key_optimizable,sq_dists_key_nonkey_edges,\ + ts_most_confident,per_nonkeyGaussian_dweight,\ + W2Cs,Depths_info_perG_nonkey, + device) + + + # keep + edges_others_optimized=edges_others + Dists_optimized2=Dists_optimizable2 + Ori_distances_optimized2=Ori_distances_optimizable2 + + return Transls_optimized2,Dists_optimized2,weight_distance_optimized2,Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others,weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict + + + + + + +def main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges,threshold_min_set): + import sys + import os + sys.path.append(os.path.dirname("../flow3d")) + + import torch + import numpy as np + from scipy.spatial.transform import Rotation + import matplotlib.pyplot as plt + import pickle + import random + import glob + + import pypose as pp + from torch import nn + + from itertools import chain + from scipy.spatial import KDTree,Delaunay,distance_matrix + import open3d as o3d + from collections import defaultdict + import roma + + # set args + + # work_dir="/data/dataset/used_ds/som/sriracha-tree/work-dir" # backpack,haru-sit, sriracha-tree + # work_dir="/data/dataset/used_ds/som/backpack/work-dir" # work-dir1 # initialization results (no SoM model) + # work_dir="/data/dataset/custom_ds_trained/rhino" + # work_dir="/data/dataset/custom_ds_graph_model/rhino" + work_dir=save_dir + + # file_effective="effective_dict.pkl" # backpack 854 + # file_effective="effective_dict_949.pkl" # haru-sit + # file_effective="effective_dict_1908.pkl" # sriracha-tree # 1908 # 1000 + # file_effective="effective_dict_484.pkl" # backpack # work-dir1 + # file_effective="effective_dict_712.pkl" # rhino + + # dict_dir=work_dir+"/out_dicts2" # has w2c and camera's K + dict_dir=work_dir+"/out_dicts3" # has w2c and camera's K + + flag_draw_contribs_sample=False + flag_visulize_sparisity=False + flag_visulize_key_point=False + flag_visulize_nonkey_point=False + + flag_auto_threshold=False + contrib_threshold=2 # usually 2 + threshold=2e-0 ## TODO # for non-key points # fixed threshold + # threshold=-torch.inf + + + + # [1] load data + nt=count_out_dict_files(dict_dir) + dicts=[] + for t in range(nt): + with open(f"{dict_dir}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # print(dicts[0].keys()) + + with open(f"{dict_dir}/{file_effective}", 'rb') as file: + effective_dict = pickle.load(file) + inds_effective = effective_dict['ids_eff2'] #ids_eff1 + # Contribs_up = effective_dict['Contribs_up'] + threshold_min = effective_dict['threshold_min'] + threshold_mean = effective_dict['threshold_mean'] + Contribs_all = effective_dict['Contribs'] + print(f"threshold_min:{threshold_min}, threshold_mean:{threshold_mean}") + + # threshold_min=np.max([threshold_min-1, 0.5]) # 0.5 + # threshold_min=np.max([threshold_min-3, 0.5]) # 0.5 # paper-windmill test + # threshold_min=0.1 # test: believe nearly all points # paper-windmill + # threshold_min=0.5 # test: camel,backpack + if threshold_min_set!=-1: + print(f"set threshold_min to {threshold_min_set}.") + threshold_min=threshold_min_set + else: + print(f"use threshold_min from effective_dict: {threshold_min}.") + + # breakpoint() + if flag_auto_threshold: + print(f"set auto threshold to {threshold_min}.") + contrib_threshold=threshold_min + threshold=threshold_min + # contrib_threshold=threshold_mean + # threshold=threshold_mean + + print(f"set threshold_min to {threshold_min}.") + + device='cuda' + ids=inds_effective + # ids=np.array(inds_effective)[np.arange(0,200,6)].tolist() # quick test + # ids=np.array(inds_effective)[np.arange(0,854,2)].tolist() + + i_target=0 + i_show=10 + Transls=np.array([dicts[t]["means"][ids].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz=np.array([dicts[t]["quats"][ids].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw=Quats_wxyz[...,[1,2,3,0]] # xyzw + # Contribs=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) #### + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids] for t in range(nt)]) #### + Contribs=Contribs_all[:,ids] + # Contribs=Contribs_up[:,ids] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort=np.array([get_order(Contribs[:,i]) for i in range(Contribs.shape[1])]).transpose(1,0) + # Vars=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) + W2Cs=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + SE3s=np.concatenate([Transls,Quats_xyzw],axis=-1) # xyz xyzw (required by pypose) + SE3s=torch.tensor(SE3s,dtype=torch.float32) #,device='cuda') + Oris_xyzw=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type) + Oris=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type).Log() + lie_tensor_SE3s=pp.LieTensor(SE3s, ltype=pp.SE3_type) + Depths_info_perG=torch.stack([dicts[t]['depths_info_perG'].cpu() for t in range(nt)]) + # depth_ratio=0.001 # 0.001 + Depths_info_perG=depth_ratio*torch.ones_like(Depths_info_perG) # TEST + print(f"TEST: set Depths_info_perG={depth_ratio}*torch.ones_like(Depths_info_perG).") # test + Depths_info_perG_key=Depths_info_perG[:,ids] + # len(dicts) + # vars=Vars[:,i_target] + nk=len(ids) # num of key points + # np.histogram(vars[vars<1e0]) + print("nk:",nk) + print("dicts[0]['means'].shape:", dicts[0]["means"].shape) + + # draw + if flag_draw_contribs_sample: + # i_show=20 + # means=Transls[:,i_show] + # vars0=Vars[:,i_show] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars0,axis=1)),cmap='viridis') + + # means=Transls[:,i_target] + # vars1=Vars[:,i_target] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars1,axis=1)),cmap='viridis') + + plt.figure(figsize=(10,4)) + # plt.subplot(1,3,1) + # plt.plot(vars0,c='r') + # plt.plot(vars1,c='b',alpha=0.6) + # plt.xlabel("time frame") + # plt.ylabel("uncertainty") + # plt.ylim([0,1]) + i_show=123 + plt.subplot(1,3,2) + plt.plot(Contribs[:,i_show],c='r') + # plt.plot(Contribs[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution") + plt.xlabel("time frame") + plt.subplot(1,3,3) + plt.plot(Contribs_sort[:,i_show],c='r') + # plt.plot(Contribs_sort[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution sort") + plt.xlabel("time frame") + plt.show() + + + # [2] Graph initialization for key Gaussians + # Initializations + # 0. naive ori distance + Oris_rel=calculate_edge_relative_orientation(Oris_xyzw) + Ori_distances_naive=get_navie_ori_distances(Oris_rel) # (nk, nk) + + # 1. vertices + # 1.1 positions + Transls=torch.tensor(Transls,dtype=torch.float32) + Transls_optimizable=Transls.clone().requires_grad_(True) + # Transls_optimizable=torch.tensor(Transls,dtype=torch.float64,requires_grad=True) + # Transls_optimizable=torch.nn.Parameter(Transls_optimizable) + # condition1a=Contribs>0.4*np.max(Contribs,axis=0) # 0.707 + # # condition1b=Contribs2>contrib_threshold + # # condition1=np.logical_or(condition1a,condition1b) + # condition1=condition1a + # Mask=torch.tensor(condition1,dtype=torch.bool) + Mask=torch.tensor(Contribs>contrib_threshold,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable=Oris_xyzw.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edge distance + Dists = np.array([distance_matrix(Transls[t], Transls[t]) for t in range(nt)]) + Dists_optimizable=np.mean(Dists,axis=0) + Dists_optimizable=torch.tensor(Dists_optimizable,dtype=torch.float32,requires_grad=True) + + Ori_distances_optimizable=Ori_distances_naive.clone() + Ori_distances_optimizable=torch.tensor(Ori_distances_optimizable,dtype=torch.float32,requires_grad=True) + + # 3. edge weight + weight_distance_optimizable=1*torch.ones(Dists_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_distance_optimizable=weight_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # use the different weight + weight_ori_distance_optimizable=1*torch.ones(Ori_distances_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_ori_distance_optimizable=weight_ori_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # 4. get local edges + n_neighbors=120 + if version_key_edges=="max_contrib": + edges_graph=get_knn_edges(n_neighbors,Transls,Contribs,Transls,Contribs,contrib_threshold,flag_remove_self=True) # edge version: max contrib version + elif version_key_edges=="minmax_distance_contrib": + # edges_graph,_=get_knn_edges_others(Transls,Mask,Transls,Mask,k=n_neighbors) # edge version: minmax distance + contrib mask version + edges_graph,_=get_knn_edges_key(Transls,Mask,k=n_neighbors) + elif version_key_edges=="max_native": + edges_graph,_=get_knn_edges_naive(n_neighbors,Transls,Transls,flag_remove_self=True) # edge version: max contrib version + # breakpoint() + else: + breakpoint() + raise ValueError("version_key_edges should be max_contrib or minmax_distance_contrib or max_native") + # edges_graph.shape # [12395, 2] + weight_distance_optimizable=torch.zeros(Dists_optimizable.shape,dtype=torch.float32) + weight_distance_optimizable[edges_graph[:,0],edges_graph[:,1]]=1 #1e-5 + weight_distance_optimizable.requires_grad=True + + # [3] optimization for key Guassians + print("===========================================") + print("Start Optimization for key Guassians") + print("-------------------------------------------") + + torch.cuda.empty_cache() + Transls_optimized,Dists_optimized,weight_distance_optimized,Oris_xyzw_optimized, Ori_distances_optimized=\ + optimize_overall_with_orientation_edges(\ + edges_graph,\ + Transls, Transls_optimizable.clone(), Mask, Dists_optimizable.clone(), weight_distance_optimizable.clone(),\ + Oris_xyzw, Oris_xyzw_optimizable.clone(), Ori_distances_optimizable.clone(), weight_ori_distance_optimizable, \ + None, None, None,None,\ + W2Cs,Depths_info_perG_key) + + print("-------------------------------------------") + print("Finish Optimization for key Guassians") + print("===========================================") + + if flag_visulize_sparisity: + # breakpoint() + weight_distance_optimized[2,:10],weight_distance_optimized[:10,2] # check symetry + # look at normed weight + torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None] + data = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=50) # check dist + # np.sort(data[4])[-100:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + # (data>1e-6).sum() + np.histogram((data>1e-6).sum(axis=-1)) + # ((data>1e-6).sum(axis=-1)==3).sum() + + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + # Example 2D array + print((data>0.02).sum()) + + # Create a meshgrid for the X and Y coordinates + x = np.arange(data.shape[1]) + y = np.arange(data.shape[0]) + x, y = np.meshgrid(x, y) + + # Create a figure and a 3D axis + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the surface + ax.plot_surface(x, y, np.flip(data,axis=1), cmap='viridis') + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_yticks([0,100,200,300,400]) + ax.set_yticklabels(nk-np.arange(0,500,100)) + ax.set_zlabel('Weight Values') + # ax.set_zlim(0,0.5) + ax.set_title('Edge Weights') + + plt.show() + + + if flag_visulize_key_point: + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=Transls.shape[0]-1 # 0 35 80 100 150 + # t=33 # test backpack + # t=0 # 306 block + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=2) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=2) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + # save + # get optimized graph from normed weights + norm_weight_distance_optimized = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() + edges_graph_optimized=[] + for i in range(nk): + for j in range(nk): + if norm_weight_distance_optimized[i,j]>1e-6: + edges_graph_optimized.append([j,i]) + edges_graph_optimized=torch.tensor(edges_graph_optimized,dtype=torch.long) + + Ori_opt=Oris_xyzw_optimized + Nodes_opt=torch.concatenate([Transls_optimized.transpose(1,0),Ori_opt.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep=[[] for i in range(nk)] + # for e in edges_opt: + for e in edges_graph_optimized: + # Edges_keep[e[0]].append([e[1],e[0]]) + Edges_keep[e[1]].append([e[0],e[1]]) + + Edges_keep=[torch.tensor(Edges_keep[i],device=device) for i in range(nk)] + num_Edges_keep=[Edges_keep[i].shape for i in range(nk)] + # num_Edges_keep + + # check + torch.norm(Oris_xyzw_optimized[0],dim=-1)[:10] + + # for save + optimized_key_output={"Transls_optimized":Transls_optimized,\ + "Dists_optimized":Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized} + + # Save variables to a file + # torch.save({ + # 'dicts': dicts,\ + # 'ids': ids,\ + # 'Contribs': Contribs,\ + # 'Transls': Transls,\ + # 'Oris_xyzw': Oris_xyzw,\ + # 'Transls_optimizable': Transls_optimizable,\ + # 'Dists_optimizable': Dists_optimizable,\ + # 'weight_distance_optimizable':weight_distance_optimizable,\ + # 'Ori_distances_optimizable':Ori_distances_optimizable,\ + # 'Transls_optimized': Transls_optimized,\ + # 'Dists_optimized': Dists_optimized,\ + # "weight_distance_optimized":weight_distance_optimized,\ + # "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + # "Ori_distances_optimized":Ori_distances_optimized,\ + # "edges_graph":edges_graph,\ + # "edges_graph_optimized":edges_graph_optimized,\ + # "Nodes_opt":Nodes_opt,\ + # "Edges_keep":Edges_keep,\ + # "optimized_key_output":optimized_key_output,\ + # # "Transls_optimized1":Transls_optimized1,\ + # # "Dists_optimized1":Dists_optimized1,\ + # # "weight_distance_optimized1":weight_distance_optimized1,\ + # "device":device,\ + # "work_dir":work_dir}\ + + # , 'variables_8.pth') + + flag_read=False + if flag_read: + # Read variables from the file + checkpoint = torch.load('variables_8.pth') + dicts = checkpoint['dicts'] + ids = checkpoint['ids'] + Contribs = checkpoint['Contribs'] + Transls = checkpoint['Transls'] + Oris_xyzw = checkpoint['Oris_xyzw'] + Transls_optimized = checkpoint['Transls_optimized'] + Dists_optimized = checkpoint['Dists_optimized'] + weight_distance_optimized = checkpoint['weight_distance_optimized'] + Oris_xyzw_optimized = checkpoint['Oris_xyzw_optimized'] + Ori_distances_optimized = checkpoint['Ori_distances_optimized'] + edges_graph = checkpoint['edges_graph'] + edges_graph_optimized=checkpoint['edges_graph_optimized'] + Nodes_opt = checkpoint['Nodes_opt'] + Edges_keep = checkpoint['Edges_keep'] + optimized_key_output = checkpoint['optimized_key_output'] + # Transls_optimized1 = checkpoint['Transls_optimized1'] + # Dists_optimized1 = checkpoint['Dists_optimized1'] + # weight_distance_optimized1 = checkpoint['weight_distance_optimized1'] + device=checkpoint['device'] + work_dir=checkpoint['work_dir'] + nk=len(ids) + + + # [4] Graph initialization for non-key Gaussians + # 1. prepare data for non-key Gaussians + Ng=dicts[0]["means"].shape[0] # num of all Gaussians + nt=len(dicts) + # Ng=1000 + ids2=[i for i in range(Ng) if i not in ids] # ids of non-target nodes + # Contribs2_temp=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2_temp=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2_temp=Contribs_all[:,ids2] + Contribs2_temp_max=Contribs2_temp.max(axis=0) + if nt>350: + n_nonkey_max=36000 # 36000 # 150000 72000 too large memory for block teddy wheel + elif nt>40: + n_nonkey_max=72000 + else: + n_nonkey_max=150000 + if Contribs2_temp_max.shape[0]>n_nonkey_max: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[-n_nonkey_max] # the nth (12000th/24000th) largest contrib value + else: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[0] + if nt>40: # iphone davis + contribs2_nth_largest=max(contribs2_nth_largest,0.01) # 0.1 + else: # nvidia + contribs2_nth_largest=max(contribs2_nth_largest,1e-4) + threshold=contribs2_nth_largest # rewrite threshold + ids2=np.array(ids2)[Contribs2_temp.max(axis=0)>threshold].tolist() # remove nodes with small contribs #15 + + + Transls2=np.array([dicts[t]["means"][ids2].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz2=np.array([dicts[t]["quats"][ids2].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw2=Quats_wxyz2[...,[1,2,3,0]] # xyzw + # Contribs2=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2=Contribs_all[:,ids2] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort2=np.array([get_order(Contribs2[:,i]) for i in range(Contribs2.shape[1])]).transpose(1,0) + # Vars2=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # W2Cs2=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + # SE3s2=np.concatenate([Transls2,Quats_xyzw2],axis=-1) # xyz xyzw (required by pypose) + # SE3s2=torch.tensor(SE3s2,dtype=torch.float32) #,device='cuda') + Oris_xyzw2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type) + Oris2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type).Log() + Depths_info_perG_nonkey=Depths_info_perG[:,ids2] + nu=len(ids2) # num of non-target nodes + print(f'Ng={Ng},nk=len(ids)={len(ids)},nu=len(ids2)={len(ids2)},threshold={threshold}') + + # # test threshold + # condition1=Contribs2>0.707*np.max(Contribs2,axis=0) + # condition2=Contribs2>contrib_threshold + # np.histogram((np.logical_or(condition1,condition1)).sum(axis=0),bins=30) + + # 2. run optimizaiton for non-key Guassians + print("===========================================") + print("Start Optimization for nonkey Guassians") + print("-------------------------------------------") + + Transls_optimized2,Dists_optimized2,weight_distance_optimized2,\ + Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others, \ + weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict \ + =run_batch(Transls2,Oris_xyzw2,Contribs2,\ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + Contribs,contrib_threshold,\ + W2Cs,Depths_info_perG_nonkey,\ + Mask) + + print("weight_param_key_optimized2 hist:",np.histogram(weight_param_key_optimized2.cpu().detach().numpy(),bins=10)) + + print("Finish Optimization for nonkey Guassians") + print("===========================================") + + + + if flag_visulize_nonkey_point: + # 3. check + # just for Method 1 + data = (torch.abs(weight_distance_optimized2)/torch.norm(weight_distance_optimized2,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=5) # check dist + np.sort(data[4])[-5:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + np.histogram((data>1e-6).sum(axis=-1)) # check sparsity + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=nt-1 # 0 35 80 100 150 + # t = 168 # backpack test + # t = 0 # 306 block # 33 + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + pts_others=Transls_optimized2.cpu().detach().numpy()[t] + # pts_others=Transls2.cpu().detach().numpy()[t] + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + x3 = pts_others[:,0] + y3 = pts_others[:,1] + z3 = pts_others[:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=4) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=4) + + ax.scatter(x3, y3, z3, c='k', label='other Points',s=1) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + Ori_opt2=torch.tensor(Quats_xyzw2,device=device) # tempory + # Ori_opt2=Oris_xyzw_optimized2.tensor() + Nodes_opt2=torch.concatenate([Transls_optimized2.transpose(1,0),Ori_opt2.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep2=[[] for i in range(nu)] + # for e in edges_opt: + for e in edges_others_optimized: + Edges_keep2[e[0]].append([e[1],e[0]+nk]) # [key id, other id + offset (nk)] + # Edges_keep2[e[1]].append([e[0],e[1]]) + + Edges_keep2=[torch.tensor(Edges_keep2[i],device=device) for i in range(nu)] + num_Edges_keep2=[Edges_keep2[i].shape for i in range(nu)] + + # combine key and others + Nodes_opt_combine=torch.cat([Nodes_opt,Nodes_opt2],dim=0) + Edges_keep_combine=Edges_keep+Edges_keep2 + ids_combine=ids+ids2 + + optimized_others_output={"Transls_optimized":Transls_optimized2,\ + "Dists_optimized":Dists_optimized2,\ + "weight_distance_optimized":weight_distance_optimized2,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized2, \ + "Ori_distances_optimized":Ori_distances_optimized2} + + + # [5] Save graph preprocessing data + quats_xyzw=Nodes_opt[:,:,3:] # xyzw + quats_wxyz=quats_xyzw[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt=[Nodes_opt[:,:,:3],quats_wxyz] # [[nk,nt,3], [nk,nt,4]] + + # combine + quats_xyzw_combine=Nodes_opt_combine[:,:,3:] # xyzw + quats_xyzw_combine=quats_xyzw_combine[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt_combine=[Nodes_opt_combine[:,:,:3],quats_xyzw_combine] # [[nk,nt,3], [nk,nt,4]] + + # tracks_SE3_opt[1].shape,Nodes_opt.shape,tracks_SE3_opt_combine[1].shape,Nodes_opt_combine.shape + + nk=len(Edges_keep) + adjacent_matrix=torch.zeros((nk,nk)) + for i in range(nk): + if Edges_keep[i].shape[0]==0: + continue + indices=Edges_keep[i][:,0] + adjacent_matrix[i,indices]=1 + + adjacent_matrix = torch.logical_or(adjacent_matrix.t(), adjacent_matrix).type(torch.int32) + + Edges_keep_bi=[] + for i in range(nk): + indices=torch.where(adjacent_matrix[i,:]==1)[0] + edges_keep_bi=torch.cat((indices.unsqueeze(0),i*torch.ones_like(indices).unsqueeze(0))).t() + Edges_keep_bi.append(edges_keep_bi) + + Edges_keep_bi_combine=Edges_keep_bi+Edges_keep2 + + + # Save 1: + is_target=list(range(len(ids))) + is_target_combine=list(range(len(ids_combine))) + opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + "optimized_key_output":optimized_key_output,\ + "Nodes_opt_combine":Nodes_opt_combine,"tracks_SE3_opt_combine":tracks_SE3_opt_combine,"ids_combine":np.array(ids_combine),\ + "is_target_combine":is_target_combine,"Edges_keep_combine":Edges_keep_combine,"Edges_keep_bi_combine":Edges_keep_bi_combine,\ + "optimized_others_output":optimized_others_output} + # opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + # "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + # "optimized_key_output":optimized_key_output} + import pickle + dict_dir=f"{work_dir}/opt_dicts" + os.makedirs(dict_dir, exist_ok=True) + file_name=f"opt_v8_{len(ids)}_0.pkl" + # file_name=f"opt_427_var0.002_cut0.002_2.pkl" # 854 eff file + # file_name=f"opt_854_var0.002_cut0.002_nearby.pkl" # 854 eff file + # file_name=f"opt_2544_var0.002_cut0.002.pkl" # 2544 eff file + # file_name=f"opt_2038_var0.002_cut0.002_test.pkl" # 2544 eff file + # file_name="opt_dicts_test3.pkl" + + # # save + # file=open(f'{dict_dir}/{file_name}', 'wb') + # pickle.dump(opt_dict, file) + # file.close() + # print("file_path:",f'{dict_dir}/{file_name}') + + # Save 2: + # Save variables to a file + dict_dir_graph=f"{work_dir}/graph_model/" + os.makedirs(dict_dir_graph, exist_ok=True) + # file_name_graph=f"v8_712_parameters0.pth" + file_path_graph=dict_dir_graph+file_name_graph + + torch.save({ + 'dicts': dicts,\ + 'ids': ids,\ + 'Contribs': Contribs,\ + 'Transls': Transls,\ + 'Oris_xyzw': Oris_xyzw,\ + 'Transls_optimizable': Transls_optimizable,\ + 'Dists_optimizable': Dists_optimizable,\ + 'weight_distance_optimizable':weight_distance_optimizable,\ + 'Ori_distances_optimizable':Ori_distances_optimizable,\ + 'Transls_optimized': Transls_optimized,\ + 'Dists_optimized': Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized,\ + "edges_graph": edges_graph,\ + "edges_graph_optimized": edges_graph_optimized,\ + "Nodes_opt":Nodes_opt,\ + "Edges_keep":Edges_keep,\ + "optimized_key_output":optimized_key_output,\ + # "Transls_optimized1":Transls_optimized1,\ + # "Dists_optimized1":Dists_optimized1,\ + # "weight_distance_optimized1":weight_distance_optimized1,\ + "device":device,\ + "work_dir":work_dir,\ + "ids2":ids2,\ + "Contribs2": Contribs2,\ + "Transls_optimized2":Transls_optimized2,\ + "Dists_optimized2":Dists_optimized2,\ + "weight_distance_optimized2":weight_distance_optimized2,\ + "Oris_xyzw_optimized2":Oris_xyzw_optimized2,\ + "Ori_distances_optimized2":Ori_distances_optimized2,\ + "edges_others_optimized":edges_others_optimized,\ + "edges_others":edges_others,\ + "blend_opt_dict":blend_opt_dict,\ + "contrib_threshold":contrib_threshold,\ + }\ + , file_path_graph) + + print("file_path_graph:",file_path_graph) + + +def main2(save_dir,voxel_size=0.4,n_min_effective_frames=5,ratio_key=-1): + flag_remove_outlier=False + flag_show_sample=False + flag_show_threshold=False + flag_ablation_select_pts_wo_uncertainty=False + flag_ablation_random_select=False + # ratio_key=0.08 # ablation study of threshold and selected number of pts at each step. + + + # folder_dict="/data/dataset/used_ds/som/sriracha-tree/work-dir/out_dicts2" # haru-sit, backpack, sriracha-tree + # folder_dict="/data/dataset/custom_ds_graph_model/rhino/out_dicts3" # rhino + folder_dict=f"{save_dir}/out_dicts3" + + nt=count_out_dict_files(folder_dict) + dicts=[] + for t in range(nt): + with open(f"{folder_dict}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy() for t in range(nt)]) #### (old) + Contribs=np.array([dicts[t]['contribs_strict'].cpu().numpy() for t in range(nt)]) #### + + # Safety check for empty Contribs + if Contribs.size == 0: + breakpoint() + print("Warning: Contribs array is empty, using contribs_in_mask as fallback") + + if ratio_key>0: + max_num_key_nodes=int(Contribs.shape[1]*ratio_key) + num_stage_wise_select_key_nodes=int(max_num_key_nodes/1.5) + max_num_key_nodes = min(max_num_key_nodes, 1500) # avoid used memory is too large. + max_num_key_nodes = max(max_num_key_nodes, 5) # avoid no of key nodes too few. + else: + max_num_key_nodes=1500 # preivous setting + num_stage_wise_select_key_nodes=1000 # previous setting + + print("******ratio_key: ",ratio_key, "max_num_key_nodes: ",max_num_key_nodes," num_stage_wise_select_key_nodes: ",num_stage_wise_select_key_nodes) + + + c99=np.quantile(Contribs, 0.99) + c80=np.quantile(Contribs, 0.80) + contrib_ref=Contribs[(Contribsc80)].mean() + if contrib_ref<0.5: + prod_num=0.5/contrib_ref + Contribs=Contribs*prod_num + Contribs = np.clip(Contribs, a_min=None, a_max=10) + # breakpoint() + # dicts[0].keys(),nt,dicts[0]['means'].shape + + if flag_remove_outlier: + def get_outliers_by_density(means): + X=means[np.arange(0,means.shape[0],1),:] + + # Apply HDBSCAN clustering + # clusterer = hdbscan.HDBSCAN(min_cluster_size=10, min_samples=5, allow_single_cluster=True) + # labels = clusterer.fit_predict(X) + clusterer = DBSCAN(eps=0.02, min_samples=10) ############### hyperparameter!!! + labels=clusterer.fit(X).labels_ + outlier_map=labels==-1 + + # Unique clusters + # n_clusters = len(set(labels)) - (1 if -1 in labels else 0) # Ignore noise label (-1) + # print(f"Detected Clusters: {n_clusters}") + # assert n_clusters==1 + + return outlier_map + + outlier_map=get_outliers_by_density(dicts[t]['means'].cpu().numpy()) + outlier_map.sum() + + Contribs_up=[] + for t in range(nt): + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + if flag_remove_outlier: + outlier_map=get_outliers_by_density(means) + contribs[outlier_map]=1e-10 ## + Contribs_up.append(contribs) + Contribs_up=np.array(Contribs_up) + else: + Contribs_up=None + + print("num of gaussians: ",dicts[0]['means'].shape[0]) + # 1st Gaussian selection round + # select according to contribs + selected_ids_ts=[] + for t in range(nt): + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + else: + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) # ascending order w.r.t. ids + n=contribs.shape[0] + # candidate_ids_map=sorted_indices>=n-1000 + candidates_in_raw_ids=sorted_indices[-num_stage_wise_select_key_nodes:] #1000 + # candidates_in_raw_ids=[i for i in range(n) if candidate_ids_map[i]] + + candidate_means=means[candidates_in_raw_ids] + # candidate_contribs=contribs[candidate_ids_map] + + # voxel_size = 0.4 # 0.5 # 0.4 + voxel_grid = np.floor(candidate_means / voxel_size) + _, unique_indices = np.unique(voxel_grid, axis=0, return_index=True) + selected_candidates_in_raw_ids=[candidates_in_raw_ids[i] for i in unique_indices] + if flag_ablation_random_select: + n_random_select=len(selected_candidates_in_raw_ids) + selected_candidates_in_raw_ids=np.random.choice(selected_candidates_in_raw_ids, n_random_select, replace=False).tolist() + selected_ids_ts.append(selected_candidates_in_raw_ids) + # print(f"t={t}, selected_candidates_in_raw_ids.shape={len(selected_candidates_in_raw_ids)}") + + counts=np.zeros(n) + for t in range(nt): + selected_candidates_in_raw_ids=selected_ids_ts[t] + counts[selected_candidates_in_raw_ids]+=1 + + selected_ids_effective_1frame=[i for i in range(n) if counts[i]>0] + print("#pts left after 1st selection round: ", len(selected_ids_effective_1frame)) + # counts[selected_ids_effective_1frame] + + # avoid too many point in the 1st selection round + # select pts + def select_pts(n,selected_ids_effective_1frame): + arr1=np.ones(n) + arr2=np.zeros(len(selected_ids_effective_1frame)-n) + arr=np.concatenate((arr1,arr2),axis=0) + np.random.shuffle(arr) + selected_ids_effective_1frame=np.array(selected_ids_effective_1frame)[arr.astype(bool)].tolist() + return selected_ids_effective_1frame + # if len(selected_ids_effective_1frame)>2200: + # selected_ids_effective_1frame=select_pts(2200,selected_ids_effective_1frame) + # print("#pts left after 1st selection round (adjusted): ", len(selected_ids_effective_1frame)) + + # 2nd Gaussian selection round + ## 2nd selection round: whose contribs are effective at least nt/6 frames + # n_min_effective_frames=30 # 30 + # n_min_effective_frames=int(nt/6) + # n_min_effective_frames=min(5,int(nt/6)) # use 30 to avoid key pts not find in the long video. # TODO # 5 for haru-sit + print("n_min_effective_frames: ",n_min_effective_frames) + counts_effective_frames=np.zeros(len(selected_ids_effective_1frame)) + thresholds=[] + for t in range(nt): + if t%100==0: + print(t) + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + else: + # means=dicts[t]['means'].cpu().numpy() + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) + # candidates_in_raw_ids=sorted_indices[-2000:] + threshold=contribs[sorted_indices[-num_stage_wise_select_key_nodes]] # the 2000th largest contrib # 1000 + threshold=max(threshold,0.5) # 0.5 + counts_effective_frames+=(contribs[selected_ids_effective_1frame]>=threshold).astype(int) + thresholds.append(threshold) + + threshold_min=np.min(np.array(thresholds)) + threshold_mean=np.mean(np.array(thresholds)) + print("min contrib threshold: ",threshold_min) + print("mean contrib threshold: ",threshold_mean) + print(np.histogram(counts_effective_frames)) + if flag_show_threshold: + import matplotlib.pyplot as plt + plt.figure() + plt.plot(thresholds) + plt.show() + selected_ids_eff=[selected_ids_effective_1frame[i] for i in range(len(selected_ids_effective_1frame)) if counts_effective_frames[i]>=n_min_effective_frames] + print("#pts left after 2nd selection round: ", len(selected_ids_eff)) + # counts_effective_frames[counts_effective_frames>30].shape + + # avoid too many point in the 2nd selection round + if len(selected_ids_eff)>max_num_key_nodes: # 1000 2500 # 1500 + selected_ids_eff=select_pts(max_num_key_nodes,selected_ids_eff) # 1000 2500 1500 + print("#pts left after 2nd selection round (adjusted): ", len(selected_ids_eff)) + + if flag_ablation_select_pts_wo_uncertainty: + selected_ids_eff=np.arange(Contribs.shape[1]) + selected_ids_eff=select_pts(1000,selected_ids_eff) # 1000 # 2500 1500 + print(f"flag_ablation_select_pts_wo_uncertainty={flag_ablation_select_pts_wo_uncertainty}") + print("random select key pts:", len(selected_ids_eff)) + + if flag_show_sample: + import matplotlib.pyplot as plt + # t=nt-1 + t=0 + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + x = means[selected_ids_eff][:,0] + y = means[selected_ids_eff][:,1] + z = means[selected_ids_eff][:,2] + # sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=np.log(contribs[selected_ids_eff]),cmap='viridis') + sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=contribs[selected_ids_eff],cmap='viridis') + + cbar = plt.colorbar(sc) + cbar.set_label('Contribs') + ax.set_xlabel('X Label') + ax.set_ylabel('Y Label') + ax.set_zlabel('Z Label') + ax.axis('equal') + plt.show() + + # save + file_path=f"{folder_dict}/effective_dict_{len(selected_ids_eff)}.pkl" + effective_dict={"ids_eff2":selected_ids_eff,"ids_eff1":selected_ids_effective_1frame,"Contribs_up":Contribs_up, + "threshold_min":threshold_min,"threshold_mean":threshold_mean,"Contribs":Contribs} + file=open(file_path, 'wb') + pickle.dump(effective_dict, file) + file.close() + print(file_path) + file_effective=f"effective_dict_{len(selected_ids_eff)}.pkl" + return file_effective + +def compute_knn_indices(points, k, chunk_size=1024): + """ + Memory-efficient KNN search: returns indices of k nearest neighbors for each point. + points: [N, 3] + returns: [N, k] LongTensor + """ + N = points.shape[0] + knn_indices = [] + + for i in range(0, N, chunk_size): + end = min(i + chunk_size, N) + chunk = points[i:end] # [chunk, 3] + dists = torch.cdist(chunk, points) # [chunk, N] + topk = dists.topk(k + 1, largest=False).indices[:, 1:] # exclude self + knn_indices.append(topk) + + return torch.cat(knn_indices, dim=0) # [N, k] + +def gaussian_kde_density_knn(points, inv_covariances, k=1000, eps=1e-6): + """ + Vectorized version using k nearest neighbors for KDE density. + points: [N, 3] + covariances: [N, 3, 3] + returns: [N] density using k-NN Mahalanobis distances + """ + N, D = points.shape + device = points.device + + # Invert covariances: [N, 3, 3] + # cov_inv = torch.inverse(covariances + eps * torch.eye(D, device=device)) + cov_inv = inv_covariances + + # # Get k-NN indices using Euclidean distance + # dists = torch.cdist(points, points) # [N, N] + # knn_indices = dists.topk(k + 1, largest=False).indices[:, 1:] # [N, k] (exclude self) + + # Memory-efficient KNN + chunk_size= 10000 + knn_indices = compute_knn_indices(points, k, chunk_size=chunk_size) # [N, k] + + # Gather neighbor points: [N, k, 3] + knn_points = points[knn_indices] # batched gather + + # Center point per row: [N, 1, 3] -> broadcast to [N, k, 3] + diffs = knn_points - points[:, None, :] # [N, k, 3] + + # Apply Mahalanobis: dᵢⱼ = (xᵢ - xⱼ)^T Σᵢ⁻¹ (xᵢ - xⱼ) + # cov_inv: [N, 3, 3], diffs: [N, k, 3] + # compute: diffᵢⱼ^T @ cov_invᵢ @ diffᵢⱼ -> result [N, k] + + diffs_unsq = diffs.unsqueeze(-1) # [N, k, 3, 1] + cov_inv_exp = cov_inv[:, None, :, :] # [N, 1, 3, 3] + m = torch.matmul(cov_inv_exp, diffs_unsq) # [N, k, 3, 1] + mahalanobis_sq = torch.matmul(diffs_unsq.transpose(-1, -2), m).squeeze(-1).squeeze(-1) # [N, k] + + # Gaussian kernel weights + weights = torch.exp(-0.5 * mahalanobis_sq) # [N, k] + + density = weights.sum(dim=1) # [N] + return density + + +def plot_density_histogram_rgb(density, bins=20, height=256): + """ + density: [N] tensor + height: desired output image height in pixels + Returns: [height, width, 3] RGB uint8 numpy image of the histogram with log-scale x-axis and labeled axes + """ + density_np = density.cpu().numpy() + density_np = density_np[density_np > 0] # filter out non-positive for log scale + + dpi = 100 # Dots per inch + fig_height_inches = height / dpi + aspect_ratio = 1.5 # width / height + fig_width_inches = fig_height_inches * aspect_ratio + + # Create figure with desired size + fig, ax = plt.subplots(figsize=(fig_width_inches, fig_height_inches), dpi=dpi) + ax.hist(density_np, bins=bins, color='skyblue', edgecolor='black') + ax.set_xscale("log") + ax.set_title("Density Histogram (Log X-axis)") + ax.set_xlabel("Density (log scale)") + ax.set_ylabel("Count") + + fig.tight_layout() + fig.canvas.draw() + + width, height = fig.canvas.get_width_height() + image = np.frombuffer(fig.canvas.tostring_rgb(), dtype=np.uint8).reshape(height, width, 3) + + plt.close(fig) + return image.astype(np.uint8) + +def load_gt_masks(ws, gt_mask_dirname="mask"): + import imageio.v2 as imageio + """Load ground truth masks from directory""" + gt_mask_dir = osp.join(ws, gt_mask_dirname) + if not osp.exists(gt_mask_dir): + logging.warning(f"GT mask directory {gt_mask_dir} not found") + breakpoint() + return None + + gt_mask_fns = [ + f for f in os.listdir(gt_mask_dir) + if f.endswith(".png") or f.endswith(".jpg") + ] + gt_mask_fns.sort() + + gt_masks = [] + for gt_mask_fn in gt_mask_fns: + gt_mask_path = osp.join(gt_mask_dir, gt_mask_fn) + gt_mask = imageio.imread(gt_mask_path) + if gt_mask.ndim == 3: + gt_mask = gt_mask[..., 0] # Take first channel if RGB + gt_mask = (gt_mask > 128).astype(np.float32) # Convert to binary + gt_masks.append(gt_mask) + + gt_masks = torch.Tensor(np.stack(gt_masks)).float() # T,H,W + # self.register_gradfree_buffer("gt_masks", gt_masks) + # self.has_gt_masks = True + logging.info(f"Loaded GT masks from {gt_mask_dir}: {gt_masks.shape}, foreground ratio: {gt_masks.mean():.3f}") + return gt_masks, gt_mask_fns + +# @torch.inference_mode() +def main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model=None): + import torch + import argparse + import os + import torch + import time + from leo_renderer import Renderer + from lib_render.render_helper import render, render_gsplat + import numpy as np + from tqdm import tqdm + import imageio.v3 as iio + + from datetime import datetime + from loguru import logger as guru + import torch.nn.functional as F + import cv2 + import matplotlib.pyplot as plt + import lpips + + from lib_ugraph.ugraph_utils import compute_ssim_map, compute_lpips_map + + # parser = argparse.ArgumentParser() + # parser.add_argument("--cfg_fn", type=str) + # parser.add_argument("--port", type=int, default=8899) + # parser.add_argument("--work_dir", type=str, required=True) + # args = parser.parse_args() + + + + if dataset_name=='iphone': + cfg_fn='/data/repo/MoSca/profile/iphone/iphone_fit.yaml' + elif dataset_name=='davis' or dataset_name=='davis_mask': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='nvidia': + cfg_fn='/data/repo/MoSca/profile/nvidia/nvidia_fit.yaml' + elif dataset_name=='diffusion4d': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='objaverse': + cfg_fn='/data/repo/MoSca/profile/objaverse/objaverse_fit.yaml' + else: + cfg_fn='' + breakpoint() + port=9000 + + device = torch.device("cuda") # if torch.cuda.is_available() else "cpu") + renderer = Renderer(cfg_fn, device, work_dir, port=port, bg_flag=True, fg_flag=True,load_s2d=True, ws=ws, use_ugraph=False, relative_dir_saved_mosca_model=relative_dir_saved_mosca_model) + + nt=renderer.cams.T + print("nt:",nt) + + W, H = renderer.cams.default_W, renderer.cams.default_H + assert W == renderer.s2d.W and H == renderer.s2d.H + K = renderer.cams.K(H=H,W=W) + # focal = 0.5 * H / np.tan(0.5 * camera_state.fov).item() + # K = torch.tensor( + # [[focal, 0.0, W / 2.0], [0.0, focal, H / 2.0], [0.0, 0.0, 1.0]], + # device=device, + # ) + # w2c = torch.linalg.inv( + # torch.from_numpy(camera_state.c2w.astype(np.float32)).to(self.device) + # ) + ts=torch.arange(nt,device=device) + w2cs=torch.stack([renderer.cams.T_cw(t) for t in ts],dim=0) + # with torch.inference_mode(): + # for w2c,t in tqdm(zip(w2cs,ts)): + # # w2c = renderer.cams.T_cw(t) + + # gs5 = [] + # if renderer.bg_flag: + # gs5.append(renderer.s_model()) + # if renderer.fg_flag: + # gs5.append(renderer.d_model(t)) + # render_dict, _ = render_gsplat( + # gs5, + # H, + # W, + # K=K, + # T_cw=w2c, + # bg_color=[1.0, 1.0, 1.0], + # ) + + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # # render_dict0 = render( + # # gs5, + # # H, + # # W, + # # K=K, + # # T_cw=w2c, + # # bg_color=[1.0, 1.0, 1.0], + # # ) + + # # img0 = torch.clamp(render_dict0["rgb"].permute(1, 2, 0), 0.0, 1.0) + # # img0 = (img0.cpu().numpy() * 255.0).astype(np.uint8) + # # combined = np.hstack([img0,img]) + # # iio.imwrite(f"./test/img_{t.item()}.png", combined) # check if they look the same. + # # compare_images(img, img0) + + if save_dir=='': + save_dir = work_dir + + video_dir = f"{save_dir}/videos/{datetime.now().strftime('%Y-%m-%d-%H%M%S')}" + print("*"*20) + print("*"*20) + print(f"Saving video to {video_dir}") + print("*"*20) + print("*"*20) + os.makedirs(video_dir, exist_ok=True) + + torch.backends.cuda.preferred_linalg_library("magma") + + # if dataset_name=="davis": + # prior_depths=torch.stack(train_dataset.depths).to(device) + # elif dataset_name=="iphone": + # prior_depths=train_dataset.depths.to(device) # [nt,H,W] + # else: + # # breakpoint() + # print("No prior depth!!!") + + if dataset_name=="davis" or dataset_name=="iphone" or dataset_name=="diffusion4d" or dataset_name=="objaverse" or dataset_name=="davis_mask": + thr_density_strict=0.1 #5 + thr_density_loose=0.1 #1 + thr_rgb_diff=0.3 + thr_ssim=0.01 + elif dataset_name=="nvidia": + thr_density_strict=0.0 + thr_density_loose=0.0 + thr_rgb_diff=0.6#0.6 + thr_ssim=0.001 # 0.001 + else: + breakpoint() + + try: + # gt_mask_dir = osp.join(ws, "mask") + # renderer.s2d.load_gt_masks(gt_mask_dir) + gt_masks=renderer.s2d.gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + except: + breakpoint() + # gt_masks=None + gt_masks, gt_mask_fns=load_gt_masks(ws, "mask") + gt_masks=gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + # breakpoint() + prior_depths=renderer.s2d.dep.to(device) # [nt,H,W] + # breakpoint() + prior_depths_grad, _, _=compute_depth_gradient(prior_depths) # [nt,H,W] + # depths_info = 0.05*torch.exp(-30*prior_depths_grad) # [nt,H,W] # TODO: hyperparameter + # depths_info = depths_info.clamp_min(1e-4) # set minimum value to 1e-4 + depths_info = 1e-3 * torch.ones_like(prior_depths_grad,device=device) # 1e-3 1e-5 + # depths_info = 1e-3 * torch.ones((nt,H,W),device=device) # 1e-3 1e-5 + # depths_info[prior_depths_grad <= 0.03] = 0.1 # [nt,H,W] # TODO: hyperparameter # 0.05 # TODO!!! + print("np.histogram:",np.histogram(prior_depths_grad[gt_masks].flatten().cpu().numpy(), bins=10)) + # np.histogram(depths_info.flatten().cpu().numpy(), bins=10) + depth_min = prior_depths.min().item() + def safe_quantile(tensor: torch.Tensor, q: float, max_samples: int = 100000): + """Compute quantile safely by subsampling if needed.""" + if tensor.numel() > max_samples: + indices = torch.randperm(tensor.numel(), device=tensor.device)[:max_samples] + tensor = tensor.view(-1)[indices] + return tensor.quantile(q) + depth_max = safe_quantile(prior_depths, 0.99).item() + prior_depths_grad_threshold = safe_quantile(prior_depths_grad[gt_masks], 0.90).item() + print(f"set prior_depths_grad_threshold to {prior_depths_grad_threshold}.") + # lpips_model = lpips.LPIPS(net="alex", spatial=True).to(device) + # breakpoint() + print("fg Gaussian num:", renderer.d_model.N) + print("bg Gaussian num:", renderer.s_model.N) + print("generate out_dicts_t.pkl:") + for i, (w2c, t) in enumerate(zip(tqdm(w2cs), ts)): + assert i == t.item() + # if i<98: + # continue # test + with torch.inference_mode(): + # out_dict=renderer.model.render(int(t.item()), w2c[None], K[None], img_wh) # my code + gs5 = [] + if renderer.bg_flag: + gs5.append(renderer.s_model()) + if renderer.fg_flag: + d_list_t=renderer.d_model(t) + gs5.append(d_list_t) + n_fg = renderer.d_model(t)[0].shape[0] + render_dict, out_dict = render_gsplat( + gs5, + H, + W, + K=K, + T_cw=w2c, + bg_color=[1.0, 1.0, 1.0], + n_fg=n_fg, + ) + n_all=len(gs5) + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # density filter begin + mu, frame, scale, o, sph=d_list_t + # S = torch.zeros_like(frame) + # S[:, 0, 0] = scale[:, 0] + # S[:, 1, 1] = scale[:, 1] + # S[:, 2, 2] = scale[:, 2] + # actual_covariance = frame @ (S**2) @ frame.permute(0, 2, 1) + inv_actual_covariance = frame @ torch.diag_embed(1.0 / (scale ** 2)) @ frame.transpose(1, 2) + + # test = torch.distributions.MultivariateNormal(mu, actual_covariance).log_prob(mu) + density=gaussian_kde_density_knn(mu,inv_actual_covariance, k=100) # TODO: check scale factor + # breakpoint() + large_density_Gaussian_strict=density>thr_density_strict # TODO: threshold # 5 + large_density_Gaussian_loose=density>thr_density_loose # TODO: threshold # 1 + # density filter end + + out_dict['w2c']=w2c.detach() + out_dict['K']=K + # gt_mask = renderer.s2d.dep_mask[i].to(device) # (H, W) # test # not the true mask! + gt_mask = gt_masks[i] + # breakpoint() + # out_dict['gt_mask']=gt_mask + projected = project_points(out_dict['means'], K, w2c) # [N,2] + in_mask_Gaussians=valid_and_visible(projected, gt_mask) # [N] bool + out_dict['in_mask_Gaussians']=in_mask_Gaussians + contribs=out_dict['contribs'].reshape(-1,3)[:,0].clone() + + img = out_dict["img"][0] # [H,W,3] + gt_img = renderer.s2d.rgb[t].to(device) # [H,W,3] + diff_img = F.l1_loss(img, gt_img, reduction="none").mean(dim=-1) # [H,W] + mask_diff_img = diff_img < thr_rgb_diff # [H,W] # TODO: threshold #0.1 + small_img_diff_Gaussians=valid_and_visible(projected, mask_diff_img) # [N] bool + + # ssim map begin + ssim_map, ssim_loss=compute_ssim_map(img.cpu().numpy(),gt_img.cpu().numpy()) # [H,W,3] + mask_ssim_img = torch.tensor(np.array(ssim_map > thr_ssim),device=device).all(dim=-1) # [H,W] # TODO + large_ssim_Gaussians=valid_and_visible(projected, mask_ssim_img) # [N] bool + + # lpips_map, lpips_loss=compute_lpips_map(lpips_model, img.cpu().numpy(),gt_img.cpu().numpy(),device=device) # [H,W,1] + # mask_lpips_img = torch.tensor((lpips_map < 0.5)[:,:,0],device=device) # [H,W] # TODO + # small_lpips_Gaussians=valid_and_visible(projected, mask_lpips_img) # [N] bool + # breakpoint() + # ssim map end + # depth gradient mask + mask_depth_grad_img = prior_depths_grad[i] < prior_depths_grad_threshold # [H,W] + small_depth_grad_Gaussians=valid_and_visible(projected, mask_depth_grad_img) # [N] bool + + # + valid_gaussians_strict = small_depth_grad_Gaussians & large_ssim_Gaussians & small_img_diff_Gaussians & in_mask_Gaussians # & large_density_Gaussian_strict #& small_lpips_Gaussians # & in_mask_Gaussians # + valid_gaussians_loose = large_ssim_Gaussians & small_img_diff_Gaussians # & large_density_Gaussian_loose #& small_lpips_Gaussians # + # valid_gaussians=torch.ones(n_all, dtype=torch.bool, device=device) # TODO: threshold + # contribs_in_mask[~small_img_diff_Gaussians]=1e-8 # Gaussians on incorrect pixel are set to small contribs + contribs_in_mask=contribs.clone() + contribs_in_mask[~in_mask_Gaussians]=1e-8 # out-mask Gaussians are set to small contribs + contribs_strict=contribs.clone() + contribs_loose=contribs.clone() + contribs_strict[~valid_gaussians_strict]=1e-8 + contribs_loose[~valid_gaussians_loose]=1e-8 + + out_dict['contribs_in_mask']=contribs_in_mask + out_dict['contribs_strict']=contribs_strict + out_dict['contribs_loose']=contribs_loose + + # depth uncertainty + depths_info_perG=associate_pts_value(projected, depths_info[i]) # per Gaussian [N] + out_dict['depths_info_perG']=depths_info_perG + # out_dict['prior_depths_grad']=prior_depths_grad + + if True: + import pickle + dict_dir=f"{save_dir}/out_dicts3" + os.makedirs(dict_dir, exist_ok=True) + file=open(f'{dict_dir}/out_dict_{t}.pkl', 'wb') + pickle.dump(out_dict, file) + file.close() + + if True: + def to_rgb(img: np.ndarray) -> np.ndarray: + # [H, W] or [H, W, 1] -> [H, W, 3] + if img.ndim == 2: + return cv2.cvtColor(img, cv2.COLOR_GRAY2BGR) + elif img.ndim == 3 and img.shape[2] == 1: + return cv2.cvtColor(img[:, :, 0], cv2.COLOR_GRAY2BGR) + return img # Already RGB or unsupported format + + img_np = (img.cpu().numpy() * 255.0).astype(np.uint8) + gt_img_np = (gt_img.cpu().numpy() * 255.0).astype(np.uint8) + ssim_map_np = (np.array(ssim_map) * 255.0).clip(0, 255).astype(np.uint8) + # lpips_map_np = to_rgb((np.array(lpips_map) * 255.0).clip(0, 255).astype(np.uint8)) + + img_with_points = draw_projected_points_masked(img_np, projected, valid_gaussians_strict) # valid_gaussians, in_mask_Gaussians + img_with_points_loose = draw_projected_points_masked(img_np, projected, valid_gaussians_loose) + masked_img = overlay_mask_yellow(img_np, gt_mask.cpu().numpy()) + masked_gt_img = overlay_mask_yellow(gt_img_np, gt_mask.cpu().numpy()) + masked_diff_gt_img = overlay_mask_yellow(gt_img_np, mask_diff_img.cpu().numpy(), color= (255,0,0)) + + diff_img_np = ((diff_img*5).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # *10 to make it clear + diff_img_np = to_rgb(diff_img_np) + # depth_info_np= ((100*depths_info[t]).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # TODO: auto normalize for show + # depth_info_np = to_rgb(depth_info_np) + img_with_points_depth=draw_projected_points_colored(img_np, projected, depths_info_perG) + depth_img=apply_depth_colormap( + prior_depths[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = to_rgb(depth_img_np) + prior_depths_grad_img=apply_depth_colormap( + prior_depths_grad[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + prior_depths_grad_img_np = (prior_depths_grad_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + + # hist_img = plot_density_histogram_rgb(density,height=img_np.shape[0]) + + # combined = stack_images_side_by_side(img_with_points, masked_img) + # lpips_map_np, masked_img, masked_gt_img, hist_img , depth_info_np, depth_img_np, + combined = np.hstack([img_with_points, img_with_points_loose, depth_img_np, prior_depths_grad_img_np, ssim_map_np, masked_img, masked_gt_img, masked_diff_gt_img, diff_img_np, img_with_points_depth]) + iio.imwrite(f"{video_dir}/img_{i}.png", combined) + # breakpoint() + # del lpips_model.net + # lpips_model=None + # Delete large objects manually + del renderer + del render_dict + del inv_actual_covariance + # del lpips_model + del prior_depths + del prior_depths_grad + out_dict = {k: v.detach().cpu() if isinstance(v, torch.Tensor) else v for k, v in out_dict.items()} + del out_dict + del img, gt_img, diff_img, mask_diff_img + del projected, depths_info_perG + del contribs_strict, contribs_loose + del ssim_map, ssim_loss#, lpips_map, lpips_loss + + + # Optionally clear lists + gs5.clear() + + # Force garbage collection + import gc + gc.collect() + + # Clear PyTorch CUDA cache + torch.cuda.empty_cache() + # breakpoint() + +def compare_images(img, img0, tolerance=1): + if img.shape != img0.shape: + print(f"Shape mismatch: {img.shape} vs {img0.shape}") + return False + + diff = np.abs(img.astype(np.int32) - img0.astype(np.int32)) + max_diff = diff.max() + + if max_diff <= tolerance: + print(f"Images are close. Max diff: {max_diff}") + return True + else: + print(f"Images differ. Max diff: {max_diff}") + return False + +# def test(): +# # iphone +# # seq_name="block" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five, spin +# seq_name="backpack_" #"backpack_" +# # work_dir=f"/data/dataset/iphone_mosca_trained/paper-windmill/logs/iphone_fit_native_add3_20250226_115418" +# work_dir=f"/data/dataset/iphone_mosca_trained/backpack_/logs/iphone_fit_native_add3" #_20250223_123930" +# save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" +# dataset_name="iphone" + +# file_name_graph=f"v12_parameters8mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) +# main3(save_dir,file_effective,file_name_graph) + +# def test3(): +# # davis +# seq_name="camel" # rhino, camel, car-turn +# work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" +# save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" +# # root_dir=f"/data/dataset/custom_mocsa_initialized" +# dataset_name="davis" + +# file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 +# main3(save_dir,file_effective,file_name_graph) + + +def test2(): + # iphone + # seq_names=['paper-windmill', 'block', 'teddy', 'apple', 'haru-sit']#, 'sriracha-tree', 'mochi-high-five', ] #,'sriracha-tree','block','teddy'] + # seq_name="mochi-high-five" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five + # 'backpack_', + # seq_names=['backpack_', 'apple', 'block', 'handwavy_', 'mochi-high-five_', 'pillow_', 'spin', 'teddy', 'creeper_', 'haru-sit_', 'paper-windmill', 'space-out', 'sriracha-tree_', 'wheel'] + # seq_names=['apple', 'block', 'spin', 'teddy', 'space-out', 'wheel'] + seq_names=['apple'] + + depth_ratio=0.1 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + # auto find the folder with timestamp + # base_dir = f"/data/dataset/iphone_mosca_trained/{seq_name}/logs" + # pattern = os.path.join(base_dir, "iphone_fit_native_add3*") + # matches = glob.glob(pattern) + # if matches: + # work_dir = matches[0] # or choose with more logic + # print("Found:", work_dir) + # else: + # print("No matching folders found.") + # print('='*20) + # continue + + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/logs/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + dataset_name="iphone" + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + # main1(dataset_name,work_dir,save_dir,ws) + torch.cuda.empty_cache() + file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="max_contrib",threshold_min=0.5) #"max_contrib" minmax_distance_contrib + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + + + +def test4(): + # davis # 'car-turn', 'camel', + # seq_names=['car-turn', 'bear', 'breakdance-flare', 'dance-twirl', 'drift-straight', 'helicopter', 'libby', 'mbike-santa', 'rhino', 'schoolgirls'] # + seq_names=['camel'] + # seq_names=['car-turn', 'bear', 'drift-straight', 'helicopter', 'mbike-santa', 'rhino'] # + depth_ratio=0.001 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + dataset_name="davis" + + file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning + # main1(dataset_name,work_dir,save_dir,ws) + file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="minmax_distance_contrib",threshold_min=0.5) #"minmax_distance_contrib") + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + +def backup_code(save_dir): + from datetime import datetime + backup_dir = osp.join(save_dir, "src_backup") + os.makedirs(backup_dir, exist_ok=True) + for path in [ + "run_preprocessing.sh", + "graph_model_preprocessing5.py", + ]: + os.system(f"cp -r {path} {backup_dir}") + # reduce the backup size + # shutil.rmtree(osp.join(backup_dir, "lib_prior", "seg")) + # for root, dirs, files in os.walk(backup_dir): + # for file in files: + # if file.endswith(".pth") or file.endswith(".ckpt"): + # if osp.isfile(osp.join(root, file)): + # os.remove(osp.join(root, file)) + # else: + # shutil.rmtree(osp.join(root, file)) + # backu the commandline args + with open(osp.join(save_dir, f"commandline_args{datetime.now().strftime('%Y-%m-%d-%H%M%S')}.txt"), "w") as f: + f.write(" ".join(sys.argv)) + +def run_from_outside(): + parser = argparse.ArgumentParser() + parser.add_argument("--seq_name", type=str, required=True) + parser.add_argument("--func_name", type=str, required=True) + parser.add_argument("--dataset_name", type=str, required=True) + parser.add_argument("--depth_ratio", type=float, help="default=0.1 (iphone) or 0.001 (backpack or davis)", required=True) # default=0.1, + parser.add_argument("--threshold_min_set", type=float, help="default=0.5 or 0.1 (paper-windmill)", required=True) # default=0.5, + parser.add_argument("--version_key_edges", type=str, help="version_key_edges: max_contrib or minmax_distance_contrib", required=True) # default="max_contrib" + parser.add_argument("--extra_save_str", type=str, help="extra_save_str", required=False, default="") # _ab_uknn ## ablation study + parser.add_argument("--ratio_key", type=float, help="ratio of key nodes to determine the selected number of nodes in the first stage and threshold in the 2nd stage", required=False, default=-1) + parser.add_argument("--which_logs_folder", type=str, help="which_logs_folder", required=False, default="logs") # default="logs" + parser.add_argument("--log_subfolder", type=str, help="log_subfolder", required=False, default="demo_fit_native_add3") # default="demo_fit_native_add3" + # parser.add_argument("--use_ugraph", action="store_true") + parser.add_argument("--relative_dir_saved_mosca_model", type=str, help="relative_dir_saved_mosca_model. e.g.,", required=False) + args = parser.parse_args() + + seq_name=args.seq_name + dataset_name=args.dataset_name + depth_ratio=args.depth_ratio # 0.001 + threshold_min_set=args.threshold_min_set # 0.5 # paper-windmill 0.1 + version_key_edges=args.version_key_edges + extra_save_str=args.extra_save_str + ratio_key=args.ratio_key + # save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}/' # original + save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}{extra_save_str}/' + logs_folder=args.which_logs_folder + relative_dir_saved_mosca_model=args.relative_dir_saved_mosca_model + log_subfolder=args.log_subfolder + if dataset_name=="iphone": + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/{logs_folder}/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + voxel_size=0.2 + n_min_effective_frames=5 + # version_key_edges="max_contrib" + elif dataset_name=="davis": + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # version_key_edges="minmax_distance_contrib" + elif dataset_name=="nvidia": + work_dir=f"/data/dataset/nvidia_mosca_trained/{seq_name}/{logs_folder}/nvidia_fit_native_add3" + save_dir=f"/data/dataset/nvidia_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/nvidia_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=3 + elif dataset_name=="diffusion4d": + work_dir=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/diffusion4d_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # elif dataset_name=="objaverse": # old + # group_name="000-001" + # # /data/dataset/objaverse_mosca_trained/000-001/3ae0b4e6a75a41d38a2adf8db048a703/000/mono/logs/demo_fit_native_add3 + # work_dir=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/{logs_folder}/demo_fit_native_add3" + # save_dir=f"/data/dataset/objaverse_mosca_graph_model/{group_name}/{seq_name}/000/mono/"+save_sub_folder + # ws=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/" + # voxel_size=0.05 + # n_min_effective_frames=5 + elif dataset_name=="objaverse": + work_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/{logs_folder}/{log_subfolder}" # TODO demo_fit_native_add3_0 + save_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_graph_model/{seq_name}/mono/"+save_sub_folder + ws=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/" + voxel_size=0.1 # TODO + n_min_effective_frames=20 # TODO + elif dataset_name=="davis_mask": + work_dir=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model_masked_4x/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + else: + raise ValueError("dataset_name not supported") + + backup_code(save_dir) + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + + if args.func_name=="main1": + main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model) + elif args.func_name=="main23": + if args.depth_ratio is None or args.threshold_min_set is None or args.version_key_edges is None: + parser.error("--depth_ratio, --threshold_min_set, and --version_key_edges are required when func_name is 'main23'") + file_effective=main2(save_dir,voxel_size=voxel_size,n_min_effective_frames=n_min_effective_frames,ratio_key=ratio_key) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges=version_key_edges,threshold_min_set=threshold_min_set) #"max_contrib" minmax_distance_contrib + + + +if __name__ == "__main__": + # test2() + # test4() + # DON'T FORGET TO CHANGE! + run_from_outside() \ No newline at end of file diff --git a/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh new file mode 100644 index 0000000000000000000000000000000000000000..fade29ad45ed691be51bd485f7dd3274bb62fad1 --- /dev/null +++ b/000-000-041ff593cec544c19a2a37cc38c06194-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh @@ -0,0 +1,159 @@ +#!/bin/bash + +gpu_id=3 +dataset_name=iphone + +# 'backpack_' 'haru-sit_' 'paper-windmill' +seq_names=('paper-windmill' 'block' 'teddy' 'wheel' 'apple' 'spin' 'space-out') +# seq_names=('teddy' 'wheel') +# seq_names=('spin' 'space-out') +# seq_names=('paper-windmill') +# seq_names=('block') +# seq_names=('haru-sit_' 'handwavy_' 'mochi-high-five_' 'sriracha-tree_' 'creeper_' 'pillow_') +# seq_names=('haru-sit_') # 'backpack_' + +depth_ratio=0.1 +# threshold_min_set=0.1 # 0.1 paper-windmill # 0.5 others +threshold_min_set_dict=( + ["paper-windmill"]=0.1 + ["teddy"]=0.5 + ["wheel"]=0.5 + ["apple"]=0.5 + ["spin"]=0.5 + ["space-out"]=0.5 + ["haru-sit_"]=0.5 + ["handwavy_"]=0.5 + ["mochi-high-five_"]=0.5 + ["sriracha-tree_"]=0.5 + ["creeper_"]=0.5 + ["pillow_"]=0.5 +) +version_key_edges=max_contrib # max_contrib minmax_distance_contrib max_native +ratio_key=0.02 # default -1 +extra_save_str=_ab_key_ratio_$ratio_key #2 + +# Loop and print each name +for seq_name in "${seq_names[@]}" +do + echo "${seq_name}" +done + +# Loop over each sequence +for seq_name in "${seq_names[@]}" +do + echo "Starting training for ${seq_name}" + + threshold_min_set=${threshold_min_set_dict[$seq_name]} + + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main1 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str # comment if no extra_save_str + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main23 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str \ + --ratio_key $ratio_key # ablation study + + echo "Finished training for ${seq_name}" + echo "*****************************************" + echo "*****************************************" +done + + + + + + +# gpu_id=2 +# dataset_name=davis +# seq_names=('camel' 'car-turn' 'drift-straight' 'helicopter' 'mbike-santa' 'koala' 'parkour' 'sheep' 'soccerball' 'schoolgirls') # 'bear' 'rhino' +# # seq_names=('camel' 'train' 'car-roundabout' 'car-turn' 'bike-packing' 'drift-straight' 'helicopter' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'mbike-santa' 'parkour' 'pigs' 'rhino' 'schoolgirls' 'sheep' 'soccerball' 'swing' 'bear') +# # seq_names=('train' 'car-roundabout' 'bike-packing' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'parkour' 'pigs' 'schoolgirls' 'sheep' 'soccerball' 'swing') + +# depth_ratio=0.001 # 0.1 +# threshold_min_set=0.1 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done + + + +# gpu_id=1 +# dataset_name=nvidia + +# # seq_names=('Balloon1' 'Balloon2' 'Jumping' 'Playground' 'Skating' 'Truck' 'Umbrella') # +# # seq_names=('Balloon1' 'Jumping' 'Truck' ) # 'Umbrella' +# seq_names=('Balloon2' 'Playground' 'Skating') + +# depth_ratio=0.1 +# threshold_min_set=0.01 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done \ No newline at end of file diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..05d1908d9f0c575de4935dc0c7d0f36469988b77 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..36ea379676b28d057d3d31de356f658a5589aade Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..1a128e697247b36b09be5f2231950e8a3e263b92 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/030/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2e1ca35cd6fbc057115cb9195c744c2b97ce6abc Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/030/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/060/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..cdb9ddc72dbd852276eb74caa3ae6bb2e72f9ddf Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/060/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-015337.txt b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-015337.txt new file mode 100644 index 0000000000000000000000000000000000000000..072063144ee0b1a25a37a1c4701d2039ca50ffb9 --- /dev/null +++ b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-015337.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-004-03638bc34bae45bbbc1ca4dfe491c767-000 --func_name main1 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-015531.txt b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-015531.txt new file mode 100644 index 0000000000000000000000000000000000000000..0d3cd0f39d0553c5b70f1479473250a186415bf8 --- /dev/null +++ b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-015531.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-004-03638bc34bae45bbbc1ca4dfe491c767-000 --func_name main23 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..cf3ab61c0adebec23ac8ed2b55d1cc23a520752c Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..25dd59f08949364a67fde87905f70c392f31fd4a Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..90462ce2a0b61617a3d31e4b5c65cbf2cb00901b Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..226215e5567ea5292c113e37f11356a7f1393176 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..b9f7a481de6336910064b7175685cd46436f34c8 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..db06931fe3bbc5f62cdf1f53a9443b5fb0bd4166 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..c7f7fdee1758b9d48241250282b2be0fd2e2bf64 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..371ae388b621e299e32040a54ed3a1fed0dec8ba Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..64a8d45b41a62b6ecf66cd8615a286dafa3b5659 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..b722f5e24af9ee2aeb82c65f11dceed4bb5e1833 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..ae011d2a2887b08d4ee8c9bd02bba3425d714e47 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..0abf26e55b93906389107a4a92773698ddb35cb5 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt new file mode 100644 index 0000000000000000000000000000000000000000..c192a7a7941da38c705af9f8497c4c762543321d --- /dev/null +++ b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt @@ -0,0 +1,6 @@ +Function Execution Times +======================== + +photometric_reconstruct_from_pretrained: 2020.2182 seconds + +Total execution time: 2020.2182 seconds diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..e4de64f67c177c01fd4eb454173383350f9eb7fc Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_120/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..a39b993c7fc100f196f0e92a8ed6f610f3ace7ba Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2b8f751df1bac4e4315bfc8c8c4b591674bc725d Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_150/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..1002d01c3e108c3fc2d36456f0d2943eecee1adb Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..4a48987cd37ab21408e699b8c0a520aa6be39899 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_210/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..08392fcbb4f3a1afb5b20ead887d4c34f162a1e9 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_240/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..1bd14e40881ce0eacbb94e5e8cf6db9b5b7c4129 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_270/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..eb1316d2875a5324d8522a0bf2b53a2181e5f334 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_30/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..e594f2c9d363eaae9ac532daec0a943fb89018bb Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_300/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2bb102e96a35740ee16a285876f0cd4121ba31a8 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_330/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..3c588b59e7a1bd7efa888eb8f5b3669702270c79 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_60/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..db368ea440341fd1b56b58f1062b481d892e4f45 Binary files /dev/null and b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/mono_90/val/val2_mask_square.xlsx differ diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt new file mode 100644 index 0000000000000000000000000000000000000000..fee54f0b89d654bc4ed725f799ff76ebd1315bf6 --- /dev/null +++ b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt @@ -0,0 +1 @@ +FPS: 162.4550032677 diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py new file mode 100644 index 0000000000000000000000000000000000000000..7016b5610ec6e7552d04c0af22d9b5b124beafb4 --- /dev/null +++ b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py @@ -0,0 +1,2928 @@ +import sys +import os +sys.path.append(os.path.dirname("../lib_render")) +sys.path.append(os.path.dirname("../lib_mosca/gs_utils")) +sys.path.append(os.path.dirname("..")) + +import torch +import numpy as np +from scipy.spatial.transform import Rotation +# import matplotlib.pyplot as plt +import pickle +import random +import glob + +# import pypose as pp +from torch import nn + +from itertools import chain +from scipy.spatial import KDTree,Delaunay,distance_matrix +import open3d as o3d +from collections import defaultdict +import roma + +from mpl_toolkits.mplot3d import Axes3D + +# import hdbscan +from sklearn.cluster import DBSCAN +from tqdm import tqdm + +from lib_mosca.mosca import __DQB_warp2__, __LBS_warp2__ + +from lib_mosca.gs_utils.utils_helper import project_points, draw_projected_points, draw_projected_points_masked, valid_and_visible, overlay_mask_yellow, stack_images_side_by_side, compute_depth_gradient, associate_pts_value, draw_projected_points_colored +from lib_mosca.gs_utils.som_vis_utils import ( + apply_depth_colormap, +) + +import matplotlib.pyplot as plt +import argparse +import logging, glob, sys, os, shutil, os.path as osp + +def get_order(arr): + """ + Computes the rank (order) of each element in the array. + + Args: + arr (np.ndarray): Input 1D array. + + Returns: + np.ndarray: Array where the i-th element is the rank of the i-th element in the input array. + """ + # Get the indices that would sort the array + sorted_indices = np.argsort(arr) + + # Create an array to store the ranks + order = np.empty_like(sorted_indices) + + # Assign ranks + order[sorted_indices] = np.arange(len(arr)) + + return order + +def count_out_dict_files(folder_path): + # Use glob to find files matching the pattern + file_pattern = os.path.join(folder_path, 'out_dict_*.pkl') + files = glob.glob(file_pattern) + return len(files) + +def calculate_edge_relative_orientation(Ts): # 854 pts will use ?? GB memory!!! + nk=Ts.shape[1] + Ts1=Ts.unsqueeze(2).repeat(1,1,nk,1) + Ts2=Ts.unsqueeze(1).repeat(1,nk,1,1) + Ts12=(Ts1.Inv()*Ts2) + # Ts_rel=relativelieAlgebra_pp(Ts1,Ts2) + return Ts12 + +def get_navie_ori_distances(Oris_rel): # TODO + Oris_rel_so3=Oris_rel.Log() # (nt, nk, nk, 3) + norm_oris_rel_so3=torch.norm(Oris_rel_so3.tensor(),dim=-1) # (nt, nk, nk) + norm_ori_rel_so3=torch.mean(norm_oris_rel_so3,axis=0) # (nk, nk) + return 1.0 - torch.cos(norm_ori_rel_so3) # (nk, nk) + + +def compute_accel_loss(transls): + # transls [nt,n,3] or oris [nt,n,4] + accel = 2 * transls[1:-1] - transls[:-2] - transls[2:] # [nt-2,n,3/4] + loss = accel.norm(dim=-1).mean() # [nt-2,n,3/4] => [nt-2,n]=> [1] + return loss +def cal_rigidity_loss(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global + stats = { + "loss_local_rigidity": loss_local_rigidity, + "loss_local_rigidity_ori": loss_local_rigidity_ori, + "loss_local_global": loss_local_global, + } + + return loss, stats + +def cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges): + oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + return loss_local_rigidity_dt,loss_local_rigidity_ori_dt + +def cal_rigidity_loss_extra(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + nt=transls.shape[0] + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + # 4. compute dt rigidity loss + # delta_t=50 + # oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + # oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + # transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + # transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + # dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + # error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # # 2. compute local rigidity loss for oris + # qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + # qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + # qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + # error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + if nt<100: + delta_t=int(nt/2) + else: + delta_t=50 + loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + # loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(50,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss_local_rigidity_10,loss_local_rigidity_ori_10=cal_dt_loss(10,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global+100*loss_local_rigidity_50+100*loss_local_rigidity_10#+100*loss_local_rigidity_ori_dt + stats = { + "loss_local_rigidity": loss_local_rigidity.item(), + "loss_local_rigidity_ori": loss_local_rigidity_ori.item(), + "loss_local_global": loss_local_global.item(), + "loss_local_rigidity_50": loss_local_rigidity_50.item(), + "loss_local_rigidity_10": loss_local_rigidity_10.item(), + } + + return loss, stats + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +# optimize_overall_with_orientation version of +def optimize_overall_with_orientation_edges( + edges, + Transls, Transls_optimizable, Mask, Dists_optimizable, weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, Ori_distances_optimizable, weight_ori_distance_optimizable, + Dists_observed_mean, Dists_observed_std, Mask_observed,Nearby_mask, + W2Cs,Depths_info_perG_key, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 ###### test + + # Depths_info_perG_key [nt,nk] + + nt = Transls.shape[0] + nk = Transls.shape[1] + ne = edges.shape[0] + + # Device setup + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + Dists_optimizable = nn.Parameter(Dists_optimizable.to(device)) + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Ori_distances_optimizable = nn.Parameter(Ori_distances_optimizable.to(device)) + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Mask = Mask.to(device) + Depths_info_perG_key=Depths_info_perG_key.to(device) + + optimizer = torch.optim.Adam( + [Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable, Ori_distances_optimizable], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, threshold=1e-3, min_lr=min_lr + ) + + # Distance matrix helper function + def distance_matrix(x, y): + return torch.cdist(x, y, p=2) + + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt,3,3] + # I0=torch.eye(3,device=device) + # I0[2,2]=0.1 # 0.01 camel + # I0=I0[None,:,:] + # I1=W2Cs_R.transpose(-1, -2)@I0@W2Cs_R # [nt,3,3] + # I1=I1[:,None,:,:] + I0=torch.eye(3,device=device).expand(nt,nk,-1,-1).clone() # [nt,nk,3,3] + I0[...,-1,-1]=Depths_info_perG_key # 0.01 camel # CHECK TODO + I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nk,3,3] + # breakpoint() + + # Optimization loop + # for epoch in range(max_epochs): + for epoch in tqdm(range(max_epochs), desc="Training Epochs", unit="epoch"): + optimizer.zero_grad() + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + a1=(Transls_optimizable-Transls)[:,:,None,:] # (nt, nk, 1, 3) + a1_T=a1.transpose(-1, -2) + squared_norm=(a1@I1@a1_T).squeeze(-1,-2) + # squared_norm=(a1@a1_T).squeeze(-1,-2) # test + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # add epsilon for numerical stability + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) # full model + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) # ablation of loss + # Dist_confident_pts_error_masked=Dist_confident_pts_error[Mask] + # Dist_confident_pts_error_masked_disturbed=Dist_confident_pts_error_masked+torch.randn_like(Dist_confident_pts_error_masked)*0.01 + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed) + # mask_shuffle=generate_ratio_binary_tensor(n=Dist_confident_pts_error_masked_disturbed.shape[0], ratio_0=0.8) + # mask_shuffle=torch.ones_like(Dist_confident_pts_error_masked_disturbed,dtype=torch.bool) + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed[mask_shuffle]) + + + # Ori_confident_pts_error=torch.norm(Oris_xyzw_optimizable_normed-Oris_xyzw,dim=-1) # test + # loss_oris_confident_pts = torch.mean(Ori_confident_pts_error[Mask]) + + mask_edge=generate_ratio_binary_tensor(n=edges.shape[0], ratio_0=0.3) + edges_mask=edges[mask_edge] + weight_normed=torch.abs(weight_distance_optimizable)/torch.norm(weight_distance_optimizable,dim=-1,p=1)[:,None] # (nu,nk) + weight_edges=weight_normed[edges_mask[:,0],edges_mask[:,1]] # (ne) + dis_local_canonical_edge=Dists_optimizable[edges_mask[:,0],edges_mask[:,1]] # (ne) CHECK order + # loss_rigidity,stats_rigidity=cal_rigidity_loss(Transls_optimizable,Oris_xyzw_optimizable_normed,edges,weight_edges,dis_local_canonical_edge) + loss_rigidity,stats_rigidity=cal_rigidity_loss_extra(Transls_optimizable,Oris_xyzw_optimizable_normed,edges_mask,weight_edges,dis_local_canonical_edge) + + # Loss 3: Regularization on weight distance + loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch[:-1] - Dists_epoch[1:])) # (nt-1,nk,nk) + # Dists_epoch_edges2=Dists_epoch_edges.reshape(nt,ne) # (nt,ne,nk) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch_edges2[:-1] - Dists_epoch_edges2[1:])) # (nt-1,nk,nk) + # loss_local_smoothness = torch.mean(torch.abs(Transls_optimizable[:-1]-Transls_optimizable[1:])) # (nt-1,nk,nk) # [TODO] + loss_accel = compute_accel_loss(Transls_optimizable) + + + # Loss 7: symmetry loss for edge weights + # TODO + # loss_symmetry=torch.mean(torch.abs(weight_distance_optimizable-weight_distance_optimizable.transpose(0,1))) + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + # a_loss_oris_confident_pts=1*loss_oris_confident_pts + a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point 0.1 common value + a_loss_sparsity=1e-1/loss_sparsity + a_loss_accel=0.1*loss_accel + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # flag_train_orientation=True + # if flag_train_orientation: + # a_loss_density=5e-3/loss_sparsity # 1e-3 for pos+rotation + # else: + # a_loss_density=5e-4/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + # a_loss_oris_distance_confident_pts=1*loss_oris_distance_confident_pts + # a_loss_ori_dist=1*loss_ori_dist + # a_loss_ori_local_smoothness=0.01*loss_ori_local_smoothness + + total_loss = a_loss_dist_confident_pts + a_loss_rigidity + a_loss_accel # + a_loss_sparsity + # + a_loss_dist + a_loss_local_smoothness + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness + a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # + a_loss_oris_confident_pts + + # Backpropagation and optimization step + total_loss.backward() + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # Logging progress + rigidity1=stats_rigidity["loss_local_rigidity"] + rigidity_ori2=stats_rigidity["loss_local_rigidity_ori"] + global3=stats_rigidity["loss_local_global"] + if epoch % log_interval == 0 or epoch == max_epochs - 1 or optimizer.param_groups[0]['lr'] < min_lr * 2: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + # f"a_loss_density: {a_loss_density.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + # f"OConfDist: {a_loss_oris_confident_pts.item():.6f}, " + f"a_ridigity: {a_loss_rigidity.item():.6f}, " + # f"rigidity1:{rigidity1:.6f}, " + # f"rigidity_ori2:{rigidity_ori2:.6f}, " + # f"global3:{global3:.6f}, " + + # f"Dist: {a_loss_dist.item():.6f}, " + # f"OConfDist {a_loss_oris_distance_confident_pts.item():.6f}, " + # f"ODist {a_loss_ori_dist.item():.6f}, " + # f"Smooth: {a_loss_local_smoothness.item():.6f}, " + # f"OSmooth: {a_loss_ori_local_smoothness.item():.6f}, " + # f"Rigidity L: {loss_local_rigidity_2.item():.6f}, " + f"a_loss_accel: {a_loss_accel.item():.6f}, " + f"a_loss_sparsity: {a_loss_sparsity.item():.6f}, " + # f"Symmetry L: {a_loss_symmetry.item():.6f}, " + # f"W L: {loss_weight_distance.item():.6f}, " + # f"W L2: {loss_weight_distance_2.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + print(f"weight_normed: {(weight_normed>0.1).sum().item():.6f}") + # print(stats_rigidity) + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and epoch > 5000: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + return Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable_normed, Ori_distances_optimizable + + +def get_knn_edges(n_neighbors,Transls,Contribs,Transls2,Contribs2,contrib_threshold, flag_remove_self=False): + # choice 2: knn graph + def o3d_knn_p(pts2,pts, num_knn, flag_remove_self=False): + indices = [] + sq_dists = [] + pcd = o3d.geometry.PointCloud() + pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + pcd_tree = o3d.geometry.KDTreeFlann(pcd) + for p in pts2: + if not flag_remove_self: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + indices.append(i[0:]) + sq_dists.append(d[0:]) + else: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn + 1) + indices.append(i[1:]) + sq_dists.append(d[1:]) + return np.array(sq_dists), np.array(indices) + + nk=Transls.shape[1] + nu=Transls2.shape[1] + + i_s_knn_others,i_s_knn_is=[],[] + # n_neighbors=5 + for i in range(nu): + t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + pts2=[pt2] # just one point + sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors, flag_remove_self) + sq_dists,indices= sq_dists[0],indices[0] + i_s_knn=i_s[indices] # usage: ids[i_s_knn] + i_s_knn_others.append(i_s_knn) + i_s_knn_is.append(i*torch.ones_like(i_s_knn)) + i_s_knn_is=torch.cat(i_s_knn_is) + i_s_knn_others=torch.cat(i_s_knn_others) + edges_others=torch.stack([i_s_knn_is,i_s_knn_others],dim=-1) + # breakpoint() + # old version: has bug when knn key points are not enough + # i_s_knn_others=torch.stack(i_s_knn_others) + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # edges_others.shape # nu*5 + return edges_others + +# def get_knn_edges_naive(n_neighbors,Transls,Transls2, flag_remove_self=False): +# nk=Transls.shape[1] +# nu=Transls2.shape[1] +# nt=Transls.shape[0] +# Dist=[] +# for t in range(nt): +# dist=(Transls2[t][:,None]-Transls[t][None]).norm(dim=-1) +# Dist.append(dist) +# Dist=torch.stack(Dist,dim=0) +# Dist=Dist.mean(dim=0) + +# def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): +# """ +# Naive temporal k-NN graph construction using average distance over time. + +# Args: +# n_neighbors (int): Number of neighbors. +# Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. +# Transls2 (torch.Tensor): (T, N2, D) tensor of query points. +# flag_remove_self (bool): If True and N1 == N2, removes self-matches. + +# Returns: +# torch.Tensor: (N2 * n_neighbors, 2) edge index pairs [Transls2_index, Transls_index]. +# """ +# T, N1, D = Transls.shape +# N2 = Transls2.shape[1] + +# # Compute temporal average pairwise distances +# dist_list = [] +# for t in range(T): +# dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) +# dist_list.append(dist) +# Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + +# if flag_remove_self and N1 == N2: +# diag = torch.arange(N2, device=Dist.device) +# Dist[diag, diag] = float('inf') + +# # Get top-k nearest neighbors for each Transls2 point +# _, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + +# # Build edge index pairs +# row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) +# edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + +# return edges_others + +def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): + """ + Naive temporal k-NN graph construction using average distance over time. + + Args: + n_neighbors (int): Number of neighbors. + Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. + Transls2 (torch.Tensor): (T, N2, D) tensor of query points. + flag_remove_self (bool): If True and N1 == N2, removes self-matches. + + Returns: + edges_others (torch.Tensor): (N2 * k, 2) edge index pairs [Transls2_index, Transls_index]. + dists (torch.Tensor): (N2 * k,) mean distances corresponding to each edge. + """ + T, N1, D = Transls.shape + N2 = Transls2.shape[1] + + # Compute temporal average pairwise distances + dist_list = [] + for t in range(T): + dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) + dist_list.append(dist) + Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + + if flag_remove_self and N1 == N2: + diag = torch.arange(N2, device=Dist.device) + Dist[diag, diag] = float('inf') + + # Get top-k nearest neighbors and distances + dists, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + + # Build edge index pairs + row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) + edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + dists_edges = dists.reshape(-1) # (N2 * k,) + + return edges_others, dists_edges + + + + + +def optimize_others_with_orientation_edges4( + edges, + Transls, Transls_optimizable, Mask, #Dists_optimizable, #weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, #Ori_distances_optimizable, #weight_ori_distance_optimizable, + Transls_key, Oris_xyzw_key,Transls_key_optimized, Oris_xyzw_key_optimized, + weight_param_key_optimizable,sq_dists_key_nonkey_edges, + ts_most_confident,per_nonkeyGaussian_dweight, + W2Cs,Depths_info_perG_nonkey, + device, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 + # weight_param_key_optimizable [nk] per-keyGuassian weight + # Depths_info_perG_nonkey [nt,nu] + # nt = ts.shape[0] + + + nt_all = Transls.shape[0] + nu=Transls.shape[1] + nk=Transls_key_optimized.shape[1] + ne=edges.shape[0] + + Oris_xyzw=Oris_xyzw/Oris_xyzw.norm(dim=-1,keepdim=True) # normalize + # Oris_xyzw_optimizable=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) # normalize + Oris_xyzw_key_optimized=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) # normalize + + # Device setup + # Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + # Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Transls_optimizable=Transls_optimizable.to(device) + Oris_xyzw_optimizable=Oris_xyzw_optimizable.tensor().to(device) + + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Transls_key = Transls_key.to(device) + Oris_xyzw_key = Oris_xyzw_key.to(device) + Transls_key_optimized = Transls_key_optimized.to(device).detach() + Oris_xyzw_key_optimized = Oris_xyzw_key_optimized.to(device).detach() + weight_param_key_optimizable = nn.Parameter(weight_param_key_optimizable.to(device)) + # sq_dists_key_nonkey = sq_dists_key_nonkey.to(device) + sq_dists_key_nonkey_edges = sq_dists_key_nonkey_edges.to(device) + per_nonkeyGaussian_dweight = nn.Parameter(per_nonkeyGaussian_dweight.to(device)) # don't use for now, just keep + # per_nonkeyGaussian_dweight=None + Depths_info_perG_nonkey = Depths_info_perG_nonkey.to(device) + + # sq_dists_key_nonkey_edges=sq_dists_key_nonkey[edges[:,0],edges[:,1]] # (ne) + exp_sq_dists_key_nonkey_edges=torch.exp(-sq_dists_key_nonkey_edges) # (ne) + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt_all,3,3] + + Mask = Mask.to(device) + + # Distance matrix helper function + # def distance_matrix(x, y): + # return torch.cdist(x, y, p=2) + + @torch.no_grad() # check + def cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu): + # weight_param_key_optimizable [nk] + # exp_sq_dists_key_nonkey_edges [ne] + device=weight_param_key_optimizable.device + nk=weight_param_key_optimizable.shape[0] + + weight_param_key_optimizable_k=100*torch.abs(weight_param_key_optimizable) # [TODO]: 2000 # 100 + weight_param_edges=weight_param_key_optimizable_k[edges[:,1]] # (ne) + # assert all(weight_param_edges>0) + # assert all(sq_dists_key_nonkey_edges>0) + # weight_edges = (exp_sq_dists_key_nonkey_edges**(1/10))**weight_param_edges # TODO: to unify scale #test: 10 for block + weight_edges = exp_sq_dists_key_nonkey_edges**weight_param_edges # TODO: to unify scale #test: 20 for block + + weight_edges = torch.clamp(weight_edges, min=1e-10, max=1e10) # test + + weights_matrix=torch.zeros((nu,nk),device=device) + weights_matrix[edges[:,0],edges[:,1]]=weight_edges + sum_temp = torch.sum(weights_matrix,dim=-1,keepdim=True) + if any(sum_temp==0): + breakpoint() + if torch.isnan(weight_param_key_optimizable).any(): + breakpoint() + weights_matrix_normed=weights_matrix/sum_temp + + weight_edges_normed=weights_matrix_normed[edges[:,0],edges[:,1]] # (ne) + + return weight_edges_normed + + + + # prepare + Oris_xyzw_key_norm=Oris_xyzw_key/Oris_xyzw_key.norm(dim=-1,keepdim=True) + Oris_xyzw_key_optimized_norm=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) + n_neighbors=(edges[:,0]==0).sum() + if n_neighbors!=5: + breakpoint() + + # new + src_xyz_1=Transls[ts_most_confident,edges[::n_neighbors,0]] # (nu,3) + src_xyzw_1=Oris_xyzw[ts_most_confident,edges[::n_neighbors,0]] # (nu,4) + src_xyz_1 = nn.Parameter(src_xyz_1) + src_xyzw_1 = nn.Parameter(src_xyzw_1) + + # get sk_src_node and sk_dst_node (key nodes) + ts_most_confident_repeat=ts_most_confident.repeat(n_neighbors,1).T.reshape(-1) # (nt*5) + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + # optimizer = torch.optim.Adam( + # [weight_param_key_optimizable,src_xyz_1,src_xyzw_1,per_nonkeyGaussian_dweight], lr=lr + # ) + optimizer = torch.optim.Adam( + [weight_param_key_optimizable,src_xyz_1,src_xyzw_1], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, min_lr=min_lr + ) + ######################################################3 + + # ts_batches = torch.split(ts_all, batch_size) + # for ts_batch in ts_batches: + # Optimization loop + # for epoch in range(max_epochs): + # ts_all=torch.arange(nt_all,device=device) + # batch_size=30 + # pbar=tqdm(range(max_epochs)) + # for step in pbar: + # optimizer.zero_grad() + ts_all = torch.arange(nt_all, device=device) + batch_size = 30 + steps_per_epoch = (nt_all + batch_size - 1) // batch_size + total_steps = steps_per_epoch * max_epochs + + pbar = tqdm(range(total_steps), desc="Training") + + for step in pbar: + epoch = step // steps_per_epoch + inner_step = step % steps_per_epoch + + # Shuffle once per epoch + if inner_step == 0: + ts_shuffled = ts_all[torch.randperm(nt_all)] + + # Select current batch + start = inner_step * batch_size + ts_batch = ts_shuffled[start:start + batch_size] + pbar.set_description(f"Epoch {epoch}, Step {inner_step}/{steps_per_epoch-1}") + + optimizer.zero_grad() + + # print(f"ts_batch: {ts_batch}") + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # Mask_DQB=Mask[ts_batch] + + sk_nt_node_xyz_1=Transls_key[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_2=sk_nt_node_xyz_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_1=Oris_xyzw_key_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_1=sk_nt_node_xyzw_1[:,:,[3,0,1,2]] + sk_nt_node_quat_2=sk_nt_node_wxyz_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,4) + sk_nt_dst_node_xyz=sk_nt_node_xyz_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat=sk_nt_node_quat_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # error propagation + I0=torch.eye(3,device=device).expand(nt,nu,-1,-1).clone() # [nt,nu,3,3] + # I0[...,-1,-1]=Depths_info_perG_nonkey # 0.001 camel # CHECK TODO + # I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nu,3,3] + I1=I0 # test # [nt,nu,3,3] + I1_masked=I1[Mask_DQB] # [n_mask,3,3] + + if torch.isnan(weight_param_key_optimizable).any(): + print("weight_param_key_optimizable is nan") + breakpoint() + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + src_xyz=src_xyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask<=nt*nu,4) + + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) # [nu,n_neighbors] + # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # sk_w=weight_per_nonkeyGaussian_adjusted.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz, # [N,K,3] + sk_nt_dst_node_quat, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + # Transls_interpolation=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + # Oris_interpolation=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_interpolation_masked=mu_dst # [n_mask,3] + Oris_interpolation_masked=quat_dst_xyzw # [n_mask,4] + + + # error propagation + # a1=(Transls_interpolation[Mask]-Transls[Mask])[:,None,:] # (n_mask, 1, 3) + a1=(Transls_interpolation_masked-Transls[ts_batch][Mask_DQB])[:,None,:] # (n_mask, 1, 3) + # a1=(Transls_interpolation_masked-Transls[ts_batch])[:,None,:] # (n_mask, 1, 3) # ablation of loss + a1_T=a1.transpose(-1, -2) # (n_mask, 3, 1) + squared_norm=(a1@I1_masked@a1_T).squeeze(-1,-2) # [n_mask] + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # [n_mask] # add epsilon for numerical stability + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) # [nt,nu] + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) + # Dist_confident_pts_error=torch.norm(Transls_interpolation[Mask]-Transls[Mask],dim=-1) # [nt,nu] # last one + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Ori_confident_pts_error=torch.norm(Oris_interpolation[Mask]-Oris_xyzw[Mask],dim=-1) # [n_mask] + Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] + # Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] # ablation of loss + loss_ori_confident_pts = torch.mean(Ori_confident_pts_error) + + # Loss 3: Regularization on weight distance + # loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_velocity = torch.mean(torch.abs(Transls_interpolation[:-1]-Transls_interpolation[1:])) # (nt-1,nk,nk) # [TODO] + # loss_ori_velocity = torch.mean(torch.abs(Oris_interpolation[:-1] - Oris_interpolation[1:])) + + # loss_accel = compute_accel_loss(Transls_interpolation) + # loss_ori_accel=compute_accel_loss(Oris_interpolation) + + + # Loss 6: regularize per nonkey Gaussian delta weight + # loss_per_nonkeyGaussian=torch.mean(torch.abs(per_nonkeyGaussian_dweight)) + # loss_per_nonkeyGaussian = 1000*torch.mean(per_nonkeyGaussian_dweight ** 2) # TODO + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + a_loss_ori_confident_pts=1*loss_ori_confident_pts + # a_loss_interpolation=1*loss_interpolation # 100 + # a_loss_interpolation_ori=1*loss_interpolation_ori # 100 + # a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point + + # a_loss_velocity=0.1*loss_velocity + # a_loss_ori_velocity=0.1*loss_ori_velocity + # a_loss_accel=0.1*loss_accel + # a_loss_ori_accel=0.1*loss_ori_accel + + # a_loss_per_nonkeyGaussian=1e-2*loss_per_nonkeyGaussian + + # a_loss_sparsity=1e-3*loss_sparsity + + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + + total_loss = a_loss_dist_confident_pts + a_loss_ori_confident_pts #+ a_loss_per_nonkeyGaussian#+ \ + # a_loss_velocity + a_loss_ori_velocity + a_loss_accel + a_loss_ori_accel #+ a_loss_sparsity \ + # a_loss_dist + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness+ a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # a_loss_rigidity + # a_loss_interpolation + a_loss_interpolation_ori + \ + + # Backpropagation and optimization step + total_loss.backward() + # total_loss.backward(retain_graph=True) + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # # Logging progress + if epoch % log_interval == 0 or epoch == max_epochs - 1: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + f"a_loss_ori_confident_pts: {a_loss_ori_confident_pts.item():.6f}, " + # f"a_loss_per_nonkeyGaussian: {a_loss_per_nonkeyGaussian.item():.6f}, " + # f"a_loss_interpolation: {a_loss_interpolation.item():.6f}, " + # f"a_loss_interpolation_ori: {a_loss_interpolation_ori.item():.6f}, " + # f"a_loss_rigidity: {a_loss_rigidity.item():.6f}, " + # f"a_loss_velocity: {a_loss_velocity.item():.6f}, " + # f"a_loss_ori_velocity: {a_loss_ori_velocity.item():.6f}, " + # f"a_loss_accel: {a_loss_accel.item():.6f}, " + # f"a_loss_ori_accel: {a_loss_ori_accel.item():.6f}, " + # f"Density L: {a_loss_density.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + # # print(f"weight_normed: {(weight_normed>1e-6).sum().item():.6f}") + # # print(f"weight_param_key_optimizable:" {weight_param_key_optimizable}) + + pbar.set_description(f"L:{total_loss.item():.6f}, "+ + f"c:{a_loss_dist_confident_pts.item():.6f}, "+ + f"oc:{a_loss_ori_confident_pts.item():.6f}, "+ + # f"per:{a_loss_per_nonkeyGaussian.item():.6f}, "+ + # f"i: {a_loss_interpolation.item():.6f}, "+ + # f"io: {a_loss_interpolation_ori.item():.6f}, "+ + # f"r:{a_loss_rigidity.item():.6f}, "+ + # f"b:{a_loss_velocity.item():.6f}, "+ + # f"ov:{a_loss_ori_velocity.item():.6f}, "+ + # f"a:{a_loss_accel.item():.6f}, "+ + # f"oa:{a_loss_ori_accel.item():.6f}, "+ + f"lr:{optimizer.param_groups[0]['lr']:.2e}") + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + # # TEMP: for keeping data structure + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + # sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) + # # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # # sk_w=weight_per_nonkeyGaussian_adjusted[None,:,:].repeat(nt,1,1)[Mask_DQB] # (n_mask,n_neighbors) + + # weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + # weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + # # for i,e in enumerate(edges): + # # weight_distance_optimizable[e[0],e[1]]=weight_edges[i] + # weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + # print("End of optimize_others_with_orientation_edges4.") + # # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + # print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + # blend_dict = { + # 'sk_w': sk_w, + # 'src_xyz': src_xyz, + # 'src_quat': src_quat, + # 'sk_nt_src_node_xyz': sk_nt_src_node_xyz, + # 'sk_nt_src_node_quat': sk_nt_src_node_quat, + # 'sk_nt_dst_node_xyz': sk_nt_dst_node_xyz, + # 'sk_nt_dst_node_quat': sk_nt_dst_node_quat + # } + + + + # ********************* new trying begin ******************* + # Step 1: calculate src_xyz_1_new and src_xyzw_1_new + weight_param_key_optimizable=weight_param_key_optimizable.abs() + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # (nu,n_neighbors) + + weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + + # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + sk_w=weight_per_nonkeyGaussian # (nu,n_neighbors) + + sk_dst_node_xyz_1=Transls_key_optimized[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_dst_node_xyzw_1=Oris_xyzw_key_optimized_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_dst_node_wxyz_1=sk_dst_node_xyzw_1[:,[3,0,1,2]] + sk_dst_node_xyz_2=sk_dst_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_dst_node_quat_2=sk_dst_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + # breakpoint() + mu_dst, quat_dst=__DQB_warp2__( # [nu,3], [nu,4] + sk_w, # [N,K] + src_xyz_1, # [N,3] + sk_src_node_xyz_2, # [nu,n_neighbors,3] + sk_src_node_quat_2, # [nu,n_neighbors,4] + sk_dst_node_xyz_2, # [N,K,3] + sk_dst_node_quat_2, # [N,K,4] + dyn_o=None, + src_quat=src_wxyz_1, # [N,4] # wxyz + ) + src_xyz_1_new=mu_dst + src_xyzw_1_new=quat_dst[:,[1,2,3,0]] # [nu,4] + # breakpoint() + # for further optimization + blend_opt_dict = { + 'weight_param_key_optimizable': weight_param_key_optimizable, # (nu,n_neighbors) + 'src_xyz_1': src_xyz_1_new, # (nu,3) # CHECK + 'src_xyzw_1': src_xyzw_1_new, # (nu,4) # CHECK + 'per_nonkeyGaussian_dweight': per_nonkeyGaussian_dweight, # (nu,n_neighbors) + 'exp_sq_dists_key_nonkey_edges': exp_sq_dists_key_nonkey_edges, # (ne) + 'ts_most_confident': ts_most_confident, # (nu) + } + + # Step 2: calculate Transls_optimized and Oris_xyzw_optimized + Transls_optimized,Oris_xyzw_optimized=[],[] + ts_batches = torch.split(ts_all, batch_size) + for ts_batch in ts_batches: + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,4) + + # key points dst needed to be updated; src is the same + sk_nt_node_xyz_op_1=Transls_key_optimized[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_op_2=sk_nt_node_xyz_op_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_op_1=Oris_xyzw_key_optimized_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_op_1=sk_nt_node_xyzw_op_1[:,:,[3,0,1,2]] + sk_nt_node_quat_op_2=sk_nt_node_wxyz_op_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_dst_node_xyz_op=sk_nt_node_xyz_op_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat_op=sk_nt_node_quat_op_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # nonkey points update when src_xyz_1 and src_xyzw_1 are trainable + src_xyz=src_xyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask<=nt*nu,4) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz_op, # [N,K,3] + sk_nt_dst_node_quat_op, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + Transls_optimized_batch=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + Oris_xyzw_optimized_batch=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_optimized.append(Transls_optimized_batch) + Oris_xyzw_optimized.append(Oris_xyzw_optimized_batch) + Transls_optimized=torch.cat(Transls_optimized,dim=0) # [nt_all,nu,3] + Oris_xyzw_optimized=torch.cat(Oris_xyzw_optimized,dim=0) # [nt_all,nu,4] + + # ********************* new trying end ******************* + # breakpoint() + blend_dict={} + return Transls_optimized, Oris_xyzw_optimized, weight_param_key_optimizable,weight_distance_optimizable,blend_dict,blend_opt_dict + +def get_knn_edges_key(Transls,Mask,k): + # group by variance of relative motion + # Transls [nt,nk,3] + # std_threshold=0.01 + # max_dist_threshold=0.4 + + def nanstd(o, dim, keepdim=False): + result = torch.sqrt( + torch.nanmean( + torch.pow( torch.abs(o-torch.nanmean(o,dim=dim).unsqueeze(dim)),2), + dim=dim + ) + ) + if keepdim: + result = result.unsqueeze(dim) + return result + + nu=Transls.shape[1] + + # Dist=torch.cdist(Transls, Transls, p=2) + Dist = torch.norm(Transls[:, :, None, :] - Transls[:, None, :, :], dim=-1) # [nt, nu, nk] + Mask_dist=Mask[:, :, None] * Mask[:, None, :] # [nt, nu, nk] + Dist_masked = torch.where(Mask_dist, Dist, torch.tensor(float('nan'), device=Dist.device)) + # std_dist=Dist.std(dim=0) # [nu,nk] + std_dist=nanstd(Dist_masked,dim=0) # [nu,nk] + + mean_dist_masked=Dist_masked.nanmean(dim=0) # [nu,nk] # may include nan + # mean_dist=Dist.mean(dim=0) # [nu,nk] + Dist_masked2 = torch.where(Mask.unsqueeze(1).expand(-1, nu, -1), Dist, torch.tensor(float('nan'), device=Dist.device)) + mean_dist_masked2=Dist_masked2.nanmean(dim=0) # [nu,nk] # should not include nan # may still have nan + # mean_dist_masked2=Dist.mean(dim=0) # [nu,nk] # overwrite [TEST] + # breakpoint() + assert torch.any(mean_dist_masked2.isnan())==False + mean_dist_masked = torch.where(~torch.isnan(std_dist), mean_dist_masked, mean_dist_masked2) + # std_dist[~torch.isnan(std_dist)].mean(),mean_dist[~torch.isnan(mean_dist)].mean() + Dist_masked2 = torch.where(Mask_dist, Dist_masked, torch.tensor(float('-inf'), device=std_dist.device)) + # np.histogram(std_dist.numpy(),bins=50) + # np.histogram(std_dist[~torch.isnan(std_dist)].cpu().numpy(), bins=50) + # std_dist.min() + + max_dist = torch.max(Dist_masked2, dim=0).values # has -inf + max_dist2 = torch.where(~torch.isinf(max_dist), max_dist, torch.tensor(float('inf'), device=max_dist.device)) + # np.histogram(max_dist2.numpy()) + # np.histogram(max_dist[~torch.isinf(max_dist)].cpu().numpy(), bins=10) + + # (std_dist<0.01).sum(),(max_dist[~torch.isinf(max_dist)]<0.4).sum(),(max_dist2<0.4).sum(),torch.isnan(std_dist).sum() + + # try 1: + # connectivity_matrix=(std_dist40: + conditiona=Contribs2>0.5*np.max(Contribs2,axis=0) # 0.707 # TODO + # conditionb=Contribs2>contrib_threshold + # condition2=np.logical_or(conditiona,conditionb) + condition2=conditiona + else: + conditionb=Contribs2>contrib_threshold + condition2=conditionb + Mask2=torch.tensor(condition2,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable2=Oris_xyzw2.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edges + # 2.1 position edges + Dists_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True) + + # 2.2 orientation edges + Ori_distances_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True)*2 + + # 3. edge weight + # 3.1 position stds + weight_distance_optimizable2=torch.zeros((nu,nk),dtype=torch.float32,requires_grad=True) + + # knn graph (max contribs version) + + # def o3d_knn_p(pts2,pts, num_knn): + # indices = [] + # sq_dists = [] + # pcd = o3d.geometry.PointCloud() + # pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + # pcd_tree = o3d.geometry.KDTreeFlann(pcd) + # for p in pts2: + # [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + # indices.append(i[0:]) + # sq_dists.append(d[0:]) + # return np.array(sq_dists), np.array(indices) + + + # i_s_knn_others=[] + # sq_dists_5=[] + # n_neighbors=5 # 5 # 10 can not work + # ts_most_confident=[] + # for i in range(nu): + # t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + # eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + # pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + # i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + # pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + # pts2=[pt2] # just one point + # sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors) + # sq_dists,indices= sq_dists[0],indices[0] + + # flag_use_key_point_key_neighbors=True + # if flag_use_key_point_key_neighbors: + # ## given a key point, get the knn n_neighbors key points + # indice_nearest_key=indices[0] # nearest key point index + # pt3=pts[indice_nearest_key] # nearest key point + # pts3=[pt3] # just one key point + # sq_dists, indices = o3d_knn_p(pts3, pts, n_neighbors) # inlcude itself + # sq_dists,indices= sq_dists[0],indices[0] + # ## + # # calcuate the distance between the i-th other point and the neighboring key points + # sq_dists=np.linalg.norm(pts[indices]-pt2,axis=-1) # [n_neighbors] + + + # i_s_knn=i_s[indices] # usage: ids[i_s_knn] + # i_s_knn_others.append(i_s_knn) + # sq_dists_5.append(sq_dists) + # ts_most_confident.append(t_most_confident) + # i_s_knn_others=torch.stack(i_s_knn_others) + # sq_dists_5=np.array(sq_dists_5) + # sq_dists_5=torch.tensor(sq_dists_5,dtype=torch.float32) + # ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # # edges_others.shape # nu*5 + + # knn graph (minmax distance + contrib version) + n_neighbors=5 + edges_others,mean_dists=get_knn_edges_others(Transls,Mask,Transls2,Mask2,k=n_neighbors) + sq_dists_5=mean_dists[edges_others[:,0],edges_others[:,1]].reshape(-1,n_neighbors) # [nu,n_neighbors] + # edges_others,dists_edges=get_knn_edges_naive(n_neighbors,Transls,Transls2,flag_remove_self=False) # ablation of uknn + # sq_dists_5=dists_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + ts_most_confident=np.argmax(Contribs2[:,:],axis=0) # [nu] + ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + torch.cuda.empty_cache() + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + + # Method 3: + # use Dual Quat Blending (refer to MoSca) + # sq_dists_key_nonkey=-torch.ones([nu,nk]) + # sq_dists_key_nonkey[edges_others[:,0],edges_others[:,1]]=sq_dists_5.reshape(-1) + sq_dists_key_nonkey_edges=sq_dists_5.reshape(-1) # [ne] + weight_param_key_optimizable=1*torch.ones(nk,dtype=torch.float32,requires_grad=True) # 2000 [TODO] + per_nonkeyGaussian_dweight=torch.zeros(nu,n_neighbors,dtype=torch.float32,requires_grad=True) + + # batch training along t + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + + Transls_optimized2,Oris_xyzw_optimized2, weight_param_key_optimized2,weight_distance_optimized2,blend_dict,blend_opt_dict=\ + optimize_others_with_orientation_edges4(\ + edges_others,\ + Transls2, Transls_optimizable2.clone(), Mask2, \ + Oris_xyzw2, Oris_xyzw_optimizable2.clone(), \ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + weight_param_key_optimizable,sq_dists_key_nonkey_edges,\ + ts_most_confident,per_nonkeyGaussian_dweight,\ + W2Cs,Depths_info_perG_nonkey, + device) + + + # keep + edges_others_optimized=edges_others + Dists_optimized2=Dists_optimizable2 + Ori_distances_optimized2=Ori_distances_optimizable2 + + return Transls_optimized2,Dists_optimized2,weight_distance_optimized2,Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others,weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict + + + + + + +def main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges,threshold_min_set): + import sys + import os + sys.path.append(os.path.dirname("../flow3d")) + + import torch + import numpy as np + from scipy.spatial.transform import Rotation + import matplotlib.pyplot as plt + import pickle + import random + import glob + + import pypose as pp + from torch import nn + + from itertools import chain + from scipy.spatial import KDTree,Delaunay,distance_matrix + import open3d as o3d + from collections import defaultdict + import roma + + # set args + + # work_dir="/data/dataset/used_ds/som/sriracha-tree/work-dir" # backpack,haru-sit, sriracha-tree + # work_dir="/data/dataset/used_ds/som/backpack/work-dir" # work-dir1 # initialization results (no SoM model) + # work_dir="/data/dataset/custom_ds_trained/rhino" + # work_dir="/data/dataset/custom_ds_graph_model/rhino" + work_dir=save_dir + + # file_effective="effective_dict.pkl" # backpack 854 + # file_effective="effective_dict_949.pkl" # haru-sit + # file_effective="effective_dict_1908.pkl" # sriracha-tree # 1908 # 1000 + # file_effective="effective_dict_484.pkl" # backpack # work-dir1 + # file_effective="effective_dict_712.pkl" # rhino + + # dict_dir=work_dir+"/out_dicts2" # has w2c and camera's K + dict_dir=work_dir+"/out_dicts3" # has w2c and camera's K + + flag_draw_contribs_sample=False + flag_visulize_sparisity=False + flag_visulize_key_point=False + flag_visulize_nonkey_point=False + + flag_auto_threshold=False + contrib_threshold=2 # usually 2 + threshold=2e-0 ## TODO # for non-key points # fixed threshold + # threshold=-torch.inf + + + + # [1] load data + nt=count_out_dict_files(dict_dir) + dicts=[] + for t in range(nt): + with open(f"{dict_dir}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # print(dicts[0].keys()) + + with open(f"{dict_dir}/{file_effective}", 'rb') as file: + effective_dict = pickle.load(file) + inds_effective = effective_dict['ids_eff2'] #ids_eff1 + # Contribs_up = effective_dict['Contribs_up'] + threshold_min = effective_dict['threshold_min'] + threshold_mean = effective_dict['threshold_mean'] + Contribs_all = effective_dict['Contribs'] + print(f"threshold_min:{threshold_min}, threshold_mean:{threshold_mean}") + + # threshold_min=np.max([threshold_min-1, 0.5]) # 0.5 + # threshold_min=np.max([threshold_min-3, 0.5]) # 0.5 # paper-windmill test + # threshold_min=0.1 # test: believe nearly all points # paper-windmill + # threshold_min=0.5 # test: camel,backpack + if threshold_min_set!=-1: + print(f"set threshold_min to {threshold_min_set}.") + threshold_min=threshold_min_set + else: + print(f"use threshold_min from effective_dict: {threshold_min}.") + + # breakpoint() + if flag_auto_threshold: + print(f"set auto threshold to {threshold_min}.") + contrib_threshold=threshold_min + threshold=threshold_min + # contrib_threshold=threshold_mean + # threshold=threshold_mean + + print(f"set threshold_min to {threshold_min}.") + + device='cuda' + ids=inds_effective + # ids=np.array(inds_effective)[np.arange(0,200,6)].tolist() # quick test + # ids=np.array(inds_effective)[np.arange(0,854,2)].tolist() + + i_target=0 + i_show=10 + Transls=np.array([dicts[t]["means"][ids].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz=np.array([dicts[t]["quats"][ids].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw=Quats_wxyz[...,[1,2,3,0]] # xyzw + # Contribs=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) #### + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids] for t in range(nt)]) #### + Contribs=Contribs_all[:,ids] + # Contribs=Contribs_up[:,ids] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort=np.array([get_order(Contribs[:,i]) for i in range(Contribs.shape[1])]).transpose(1,0) + # Vars=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) + W2Cs=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + SE3s=np.concatenate([Transls,Quats_xyzw],axis=-1) # xyz xyzw (required by pypose) + SE3s=torch.tensor(SE3s,dtype=torch.float32) #,device='cuda') + Oris_xyzw=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type) + Oris=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type).Log() + lie_tensor_SE3s=pp.LieTensor(SE3s, ltype=pp.SE3_type) + Depths_info_perG=torch.stack([dicts[t]['depths_info_perG'].cpu() for t in range(nt)]) + # depth_ratio=0.001 # 0.001 + Depths_info_perG=depth_ratio*torch.ones_like(Depths_info_perG) # TEST + print(f"TEST: set Depths_info_perG={depth_ratio}*torch.ones_like(Depths_info_perG).") # test + Depths_info_perG_key=Depths_info_perG[:,ids] + # len(dicts) + # vars=Vars[:,i_target] + nk=len(ids) # num of key points + # np.histogram(vars[vars<1e0]) + print("nk:",nk) + print("dicts[0]['means'].shape:", dicts[0]["means"].shape) + + # draw + if flag_draw_contribs_sample: + # i_show=20 + # means=Transls[:,i_show] + # vars0=Vars[:,i_show] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars0,axis=1)),cmap='viridis') + + # means=Transls[:,i_target] + # vars1=Vars[:,i_target] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars1,axis=1)),cmap='viridis') + + plt.figure(figsize=(10,4)) + # plt.subplot(1,3,1) + # plt.plot(vars0,c='r') + # plt.plot(vars1,c='b',alpha=0.6) + # plt.xlabel("time frame") + # plt.ylabel("uncertainty") + # plt.ylim([0,1]) + i_show=123 + plt.subplot(1,3,2) + plt.plot(Contribs[:,i_show],c='r') + # plt.plot(Contribs[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution") + plt.xlabel("time frame") + plt.subplot(1,3,3) + plt.plot(Contribs_sort[:,i_show],c='r') + # plt.plot(Contribs_sort[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution sort") + plt.xlabel("time frame") + plt.show() + + + # [2] Graph initialization for key Gaussians + # Initializations + # 0. naive ori distance + Oris_rel=calculate_edge_relative_orientation(Oris_xyzw) + Ori_distances_naive=get_navie_ori_distances(Oris_rel) # (nk, nk) + + # 1. vertices + # 1.1 positions + Transls=torch.tensor(Transls,dtype=torch.float32) + Transls_optimizable=Transls.clone().requires_grad_(True) + # Transls_optimizable=torch.tensor(Transls,dtype=torch.float64,requires_grad=True) + # Transls_optimizable=torch.nn.Parameter(Transls_optimizable) + # condition1a=Contribs>0.4*np.max(Contribs,axis=0) # 0.707 + # # condition1b=Contribs2>contrib_threshold + # # condition1=np.logical_or(condition1a,condition1b) + # condition1=condition1a + # Mask=torch.tensor(condition1,dtype=torch.bool) + Mask=torch.tensor(Contribs>contrib_threshold,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable=Oris_xyzw.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edge distance + Dists = np.array([distance_matrix(Transls[t], Transls[t]) for t in range(nt)]) + Dists_optimizable=np.mean(Dists,axis=0) + Dists_optimizable=torch.tensor(Dists_optimizable,dtype=torch.float32,requires_grad=True) + + Ori_distances_optimizable=Ori_distances_naive.clone() + Ori_distances_optimizable=torch.tensor(Ori_distances_optimizable,dtype=torch.float32,requires_grad=True) + + # 3. edge weight + weight_distance_optimizable=1*torch.ones(Dists_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_distance_optimizable=weight_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # use the different weight + weight_ori_distance_optimizable=1*torch.ones(Ori_distances_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_ori_distance_optimizable=weight_ori_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # 4. get local edges + n_neighbors=120 + if version_key_edges=="max_contrib": + edges_graph=get_knn_edges(n_neighbors,Transls,Contribs,Transls,Contribs,contrib_threshold,flag_remove_self=True) # edge version: max contrib version + elif version_key_edges=="minmax_distance_contrib": + # edges_graph,_=get_knn_edges_others(Transls,Mask,Transls,Mask,k=n_neighbors) # edge version: minmax distance + contrib mask version + edges_graph,_=get_knn_edges_key(Transls,Mask,k=n_neighbors) + elif version_key_edges=="max_native": + edges_graph,_=get_knn_edges_naive(n_neighbors,Transls,Transls,flag_remove_self=True) # edge version: max contrib version + # breakpoint() + else: + breakpoint() + raise ValueError("version_key_edges should be max_contrib or minmax_distance_contrib or max_native") + # edges_graph.shape # [12395, 2] + weight_distance_optimizable=torch.zeros(Dists_optimizable.shape,dtype=torch.float32) + weight_distance_optimizable[edges_graph[:,0],edges_graph[:,1]]=1 #1e-5 + weight_distance_optimizable.requires_grad=True + + # [3] optimization for key Guassians + print("===========================================") + print("Start Optimization for key Guassians") + print("-------------------------------------------") + + torch.cuda.empty_cache() + Transls_optimized,Dists_optimized,weight_distance_optimized,Oris_xyzw_optimized, Ori_distances_optimized=\ + optimize_overall_with_orientation_edges(\ + edges_graph,\ + Transls, Transls_optimizable.clone(), Mask, Dists_optimizable.clone(), weight_distance_optimizable.clone(),\ + Oris_xyzw, Oris_xyzw_optimizable.clone(), Ori_distances_optimizable.clone(), weight_ori_distance_optimizable, \ + None, None, None,None,\ + W2Cs,Depths_info_perG_key) + + print("-------------------------------------------") + print("Finish Optimization for key Guassians") + print("===========================================") + + if flag_visulize_sparisity: + # breakpoint() + weight_distance_optimized[2,:10],weight_distance_optimized[:10,2] # check symetry + # look at normed weight + torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None] + data = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=50) # check dist + # np.sort(data[4])[-100:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + # (data>1e-6).sum() + np.histogram((data>1e-6).sum(axis=-1)) + # ((data>1e-6).sum(axis=-1)==3).sum() + + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + # Example 2D array + print((data>0.02).sum()) + + # Create a meshgrid for the X and Y coordinates + x = np.arange(data.shape[1]) + y = np.arange(data.shape[0]) + x, y = np.meshgrid(x, y) + + # Create a figure and a 3D axis + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the surface + ax.plot_surface(x, y, np.flip(data,axis=1), cmap='viridis') + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_yticks([0,100,200,300,400]) + ax.set_yticklabels(nk-np.arange(0,500,100)) + ax.set_zlabel('Weight Values') + # ax.set_zlim(0,0.5) + ax.set_title('Edge Weights') + + plt.show() + + + if flag_visulize_key_point: + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=Transls.shape[0]-1 # 0 35 80 100 150 + # t=33 # test backpack + # t=0 # 306 block + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=2) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=2) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + # save + # get optimized graph from normed weights + norm_weight_distance_optimized = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() + edges_graph_optimized=[] + for i in range(nk): + for j in range(nk): + if norm_weight_distance_optimized[i,j]>1e-6: + edges_graph_optimized.append([j,i]) + edges_graph_optimized=torch.tensor(edges_graph_optimized,dtype=torch.long) + + Ori_opt=Oris_xyzw_optimized + Nodes_opt=torch.concatenate([Transls_optimized.transpose(1,0),Ori_opt.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep=[[] for i in range(nk)] + # for e in edges_opt: + for e in edges_graph_optimized: + # Edges_keep[e[0]].append([e[1],e[0]]) + Edges_keep[e[1]].append([e[0],e[1]]) + + Edges_keep=[torch.tensor(Edges_keep[i],device=device) for i in range(nk)] + num_Edges_keep=[Edges_keep[i].shape for i in range(nk)] + # num_Edges_keep + + # check + torch.norm(Oris_xyzw_optimized[0],dim=-1)[:10] + + # for save + optimized_key_output={"Transls_optimized":Transls_optimized,\ + "Dists_optimized":Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized} + + # Save variables to a file + # torch.save({ + # 'dicts': dicts,\ + # 'ids': ids,\ + # 'Contribs': Contribs,\ + # 'Transls': Transls,\ + # 'Oris_xyzw': Oris_xyzw,\ + # 'Transls_optimizable': Transls_optimizable,\ + # 'Dists_optimizable': Dists_optimizable,\ + # 'weight_distance_optimizable':weight_distance_optimizable,\ + # 'Ori_distances_optimizable':Ori_distances_optimizable,\ + # 'Transls_optimized': Transls_optimized,\ + # 'Dists_optimized': Dists_optimized,\ + # "weight_distance_optimized":weight_distance_optimized,\ + # "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + # "Ori_distances_optimized":Ori_distances_optimized,\ + # "edges_graph":edges_graph,\ + # "edges_graph_optimized":edges_graph_optimized,\ + # "Nodes_opt":Nodes_opt,\ + # "Edges_keep":Edges_keep,\ + # "optimized_key_output":optimized_key_output,\ + # # "Transls_optimized1":Transls_optimized1,\ + # # "Dists_optimized1":Dists_optimized1,\ + # # "weight_distance_optimized1":weight_distance_optimized1,\ + # "device":device,\ + # "work_dir":work_dir}\ + + # , 'variables_8.pth') + + flag_read=False + if flag_read: + # Read variables from the file + checkpoint = torch.load('variables_8.pth') + dicts = checkpoint['dicts'] + ids = checkpoint['ids'] + Contribs = checkpoint['Contribs'] + Transls = checkpoint['Transls'] + Oris_xyzw = checkpoint['Oris_xyzw'] + Transls_optimized = checkpoint['Transls_optimized'] + Dists_optimized = checkpoint['Dists_optimized'] + weight_distance_optimized = checkpoint['weight_distance_optimized'] + Oris_xyzw_optimized = checkpoint['Oris_xyzw_optimized'] + Ori_distances_optimized = checkpoint['Ori_distances_optimized'] + edges_graph = checkpoint['edges_graph'] + edges_graph_optimized=checkpoint['edges_graph_optimized'] + Nodes_opt = checkpoint['Nodes_opt'] + Edges_keep = checkpoint['Edges_keep'] + optimized_key_output = checkpoint['optimized_key_output'] + # Transls_optimized1 = checkpoint['Transls_optimized1'] + # Dists_optimized1 = checkpoint['Dists_optimized1'] + # weight_distance_optimized1 = checkpoint['weight_distance_optimized1'] + device=checkpoint['device'] + work_dir=checkpoint['work_dir'] + nk=len(ids) + + + # [4] Graph initialization for non-key Gaussians + # 1. prepare data for non-key Gaussians + Ng=dicts[0]["means"].shape[0] # num of all Gaussians + nt=len(dicts) + # Ng=1000 + ids2=[i for i in range(Ng) if i not in ids] # ids of non-target nodes + # Contribs2_temp=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2_temp=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2_temp=Contribs_all[:,ids2] + Contribs2_temp_max=Contribs2_temp.max(axis=0) + if nt>350: + n_nonkey_max=36000 # 36000 # 150000 72000 too large memory for block teddy wheel + elif nt>40: + n_nonkey_max=72000 + else: + n_nonkey_max=150000 + if Contribs2_temp_max.shape[0]>n_nonkey_max: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[-n_nonkey_max] # the nth (12000th/24000th) largest contrib value + else: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[0] + if nt>40: # iphone davis + contribs2_nth_largest=max(contribs2_nth_largest,0.01) # 0.1 + else: # nvidia + contribs2_nth_largest=max(contribs2_nth_largest,1e-4) + threshold=contribs2_nth_largest # rewrite threshold + ids2=np.array(ids2)[Contribs2_temp.max(axis=0)>threshold].tolist() # remove nodes with small contribs #15 + + + Transls2=np.array([dicts[t]["means"][ids2].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz2=np.array([dicts[t]["quats"][ids2].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw2=Quats_wxyz2[...,[1,2,3,0]] # xyzw + # Contribs2=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2=Contribs_all[:,ids2] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort2=np.array([get_order(Contribs2[:,i]) for i in range(Contribs2.shape[1])]).transpose(1,0) + # Vars2=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # W2Cs2=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + # SE3s2=np.concatenate([Transls2,Quats_xyzw2],axis=-1) # xyz xyzw (required by pypose) + # SE3s2=torch.tensor(SE3s2,dtype=torch.float32) #,device='cuda') + Oris_xyzw2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type) + Oris2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type).Log() + Depths_info_perG_nonkey=Depths_info_perG[:,ids2] + nu=len(ids2) # num of non-target nodes + print(f'Ng={Ng},nk=len(ids)={len(ids)},nu=len(ids2)={len(ids2)},threshold={threshold}') + + # # test threshold + # condition1=Contribs2>0.707*np.max(Contribs2,axis=0) + # condition2=Contribs2>contrib_threshold + # np.histogram((np.logical_or(condition1,condition1)).sum(axis=0),bins=30) + + # 2. run optimizaiton for non-key Guassians + print("===========================================") + print("Start Optimization for nonkey Guassians") + print("-------------------------------------------") + + Transls_optimized2,Dists_optimized2,weight_distance_optimized2,\ + Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others, \ + weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict \ + =run_batch(Transls2,Oris_xyzw2,Contribs2,\ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + Contribs,contrib_threshold,\ + W2Cs,Depths_info_perG_nonkey,\ + Mask) + + print("weight_param_key_optimized2 hist:",np.histogram(weight_param_key_optimized2.cpu().detach().numpy(),bins=10)) + + print("Finish Optimization for nonkey Guassians") + print("===========================================") + + + + if flag_visulize_nonkey_point: + # 3. check + # just for Method 1 + data = (torch.abs(weight_distance_optimized2)/torch.norm(weight_distance_optimized2,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=5) # check dist + np.sort(data[4])[-5:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + np.histogram((data>1e-6).sum(axis=-1)) # check sparsity + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=nt-1 # 0 35 80 100 150 + # t = 168 # backpack test + # t = 0 # 306 block # 33 + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + pts_others=Transls_optimized2.cpu().detach().numpy()[t] + # pts_others=Transls2.cpu().detach().numpy()[t] + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + x3 = pts_others[:,0] + y3 = pts_others[:,1] + z3 = pts_others[:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=4) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=4) + + ax.scatter(x3, y3, z3, c='k', label='other Points',s=1) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + Ori_opt2=torch.tensor(Quats_xyzw2,device=device) # tempory + # Ori_opt2=Oris_xyzw_optimized2.tensor() + Nodes_opt2=torch.concatenate([Transls_optimized2.transpose(1,0),Ori_opt2.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep2=[[] for i in range(nu)] + # for e in edges_opt: + for e in edges_others_optimized: + Edges_keep2[e[0]].append([e[1],e[0]+nk]) # [key id, other id + offset (nk)] + # Edges_keep2[e[1]].append([e[0],e[1]]) + + Edges_keep2=[torch.tensor(Edges_keep2[i],device=device) for i in range(nu)] + num_Edges_keep2=[Edges_keep2[i].shape for i in range(nu)] + + # combine key and others + Nodes_opt_combine=torch.cat([Nodes_opt,Nodes_opt2],dim=0) + Edges_keep_combine=Edges_keep+Edges_keep2 + ids_combine=ids+ids2 + + optimized_others_output={"Transls_optimized":Transls_optimized2,\ + "Dists_optimized":Dists_optimized2,\ + "weight_distance_optimized":weight_distance_optimized2,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized2, \ + "Ori_distances_optimized":Ori_distances_optimized2} + + + # [5] Save graph preprocessing data + quats_xyzw=Nodes_opt[:,:,3:] # xyzw + quats_wxyz=quats_xyzw[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt=[Nodes_opt[:,:,:3],quats_wxyz] # [[nk,nt,3], [nk,nt,4]] + + # combine + quats_xyzw_combine=Nodes_opt_combine[:,:,3:] # xyzw + quats_xyzw_combine=quats_xyzw_combine[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt_combine=[Nodes_opt_combine[:,:,:3],quats_xyzw_combine] # [[nk,nt,3], [nk,nt,4]] + + # tracks_SE3_opt[1].shape,Nodes_opt.shape,tracks_SE3_opt_combine[1].shape,Nodes_opt_combine.shape + + nk=len(Edges_keep) + adjacent_matrix=torch.zeros((nk,nk)) + for i in range(nk): + if Edges_keep[i].shape[0]==0: + continue + indices=Edges_keep[i][:,0] + adjacent_matrix[i,indices]=1 + + adjacent_matrix = torch.logical_or(adjacent_matrix.t(), adjacent_matrix).type(torch.int32) + + Edges_keep_bi=[] + for i in range(nk): + indices=torch.where(adjacent_matrix[i,:]==1)[0] + edges_keep_bi=torch.cat((indices.unsqueeze(0),i*torch.ones_like(indices).unsqueeze(0))).t() + Edges_keep_bi.append(edges_keep_bi) + + Edges_keep_bi_combine=Edges_keep_bi+Edges_keep2 + + + # Save 1: + is_target=list(range(len(ids))) + is_target_combine=list(range(len(ids_combine))) + opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + "optimized_key_output":optimized_key_output,\ + "Nodes_opt_combine":Nodes_opt_combine,"tracks_SE3_opt_combine":tracks_SE3_opt_combine,"ids_combine":np.array(ids_combine),\ + "is_target_combine":is_target_combine,"Edges_keep_combine":Edges_keep_combine,"Edges_keep_bi_combine":Edges_keep_bi_combine,\ + "optimized_others_output":optimized_others_output} + # opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + # "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + # "optimized_key_output":optimized_key_output} + import pickle + dict_dir=f"{work_dir}/opt_dicts" + os.makedirs(dict_dir, exist_ok=True) + file_name=f"opt_v8_{len(ids)}_0.pkl" + # file_name=f"opt_427_var0.002_cut0.002_2.pkl" # 854 eff file + # file_name=f"opt_854_var0.002_cut0.002_nearby.pkl" # 854 eff file + # file_name=f"opt_2544_var0.002_cut0.002.pkl" # 2544 eff file + # file_name=f"opt_2038_var0.002_cut0.002_test.pkl" # 2544 eff file + # file_name="opt_dicts_test3.pkl" + + # # save + # file=open(f'{dict_dir}/{file_name}', 'wb') + # pickle.dump(opt_dict, file) + # file.close() + # print("file_path:",f'{dict_dir}/{file_name}') + + # Save 2: + # Save variables to a file + dict_dir_graph=f"{work_dir}/graph_model/" + os.makedirs(dict_dir_graph, exist_ok=True) + # file_name_graph=f"v8_712_parameters0.pth" + file_path_graph=dict_dir_graph+file_name_graph + + torch.save({ + 'dicts': dicts,\ + 'ids': ids,\ + 'Contribs': Contribs,\ + 'Transls': Transls,\ + 'Oris_xyzw': Oris_xyzw,\ + 'Transls_optimizable': Transls_optimizable,\ + 'Dists_optimizable': Dists_optimizable,\ + 'weight_distance_optimizable':weight_distance_optimizable,\ + 'Ori_distances_optimizable':Ori_distances_optimizable,\ + 'Transls_optimized': Transls_optimized,\ + 'Dists_optimized': Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized,\ + "edges_graph": edges_graph,\ + "edges_graph_optimized": edges_graph_optimized,\ + "Nodes_opt":Nodes_opt,\ + "Edges_keep":Edges_keep,\ + "optimized_key_output":optimized_key_output,\ + # "Transls_optimized1":Transls_optimized1,\ + # "Dists_optimized1":Dists_optimized1,\ + # "weight_distance_optimized1":weight_distance_optimized1,\ + "device":device,\ + "work_dir":work_dir,\ + "ids2":ids2,\ + "Contribs2": Contribs2,\ + "Transls_optimized2":Transls_optimized2,\ + "Dists_optimized2":Dists_optimized2,\ + "weight_distance_optimized2":weight_distance_optimized2,\ + "Oris_xyzw_optimized2":Oris_xyzw_optimized2,\ + "Ori_distances_optimized2":Ori_distances_optimized2,\ + "edges_others_optimized":edges_others_optimized,\ + "edges_others":edges_others,\ + "blend_opt_dict":blend_opt_dict,\ + "contrib_threshold":contrib_threshold,\ + }\ + , file_path_graph) + + print("file_path_graph:",file_path_graph) + + +def main2(save_dir,voxel_size=0.4,n_min_effective_frames=5,ratio_key=-1): + flag_remove_outlier=False + flag_show_sample=False + flag_show_threshold=False + flag_ablation_select_pts_wo_uncertainty=False + flag_ablation_random_select=False + # ratio_key=0.08 # ablation study of threshold and selected number of pts at each step. + + + # folder_dict="/data/dataset/used_ds/som/sriracha-tree/work-dir/out_dicts2" # haru-sit, backpack, sriracha-tree + # folder_dict="/data/dataset/custom_ds_graph_model/rhino/out_dicts3" # rhino + folder_dict=f"{save_dir}/out_dicts3" + + nt=count_out_dict_files(folder_dict) + dicts=[] + for t in range(nt): + with open(f"{folder_dict}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy() for t in range(nt)]) #### (old) + Contribs=np.array([dicts[t]['contribs_strict'].cpu().numpy() for t in range(nt)]) #### + + # Safety check for empty Contribs + if Contribs.size == 0: + breakpoint() + print("Warning: Contribs array is empty, using contribs_in_mask as fallback") + + if ratio_key>0: + max_num_key_nodes=int(Contribs.shape[1]*ratio_key) + num_stage_wise_select_key_nodes=int(max_num_key_nodes/1.5) + max_num_key_nodes = min(max_num_key_nodes, 1500) # avoid used memory is too large. + max_num_key_nodes = max(max_num_key_nodes, 5) # avoid no of key nodes too few. + else: + max_num_key_nodes=1500 # preivous setting + num_stage_wise_select_key_nodes=1000 # previous setting + + print("******ratio_key: ",ratio_key, "max_num_key_nodes: ",max_num_key_nodes," num_stage_wise_select_key_nodes: ",num_stage_wise_select_key_nodes) + + + c99=np.quantile(Contribs, 0.99) + c80=np.quantile(Contribs, 0.80) + contrib_ref=Contribs[(Contribsc80)].mean() + if contrib_ref<0.5: + prod_num=0.5/contrib_ref + Contribs=Contribs*prod_num + Contribs = np.clip(Contribs, a_min=None, a_max=10) + # breakpoint() + # dicts[0].keys(),nt,dicts[0]['means'].shape + + if flag_remove_outlier: + def get_outliers_by_density(means): + X=means[np.arange(0,means.shape[0],1),:] + + # Apply HDBSCAN clustering + # clusterer = hdbscan.HDBSCAN(min_cluster_size=10, min_samples=5, allow_single_cluster=True) + # labels = clusterer.fit_predict(X) + clusterer = DBSCAN(eps=0.02, min_samples=10) ############### hyperparameter!!! + labels=clusterer.fit(X).labels_ + outlier_map=labels==-1 + + # Unique clusters + # n_clusters = len(set(labels)) - (1 if -1 in labels else 0) # Ignore noise label (-1) + # print(f"Detected Clusters: {n_clusters}") + # assert n_clusters==1 + + return outlier_map + + outlier_map=get_outliers_by_density(dicts[t]['means'].cpu().numpy()) + outlier_map.sum() + + Contribs_up=[] + for t in range(nt): + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + if flag_remove_outlier: + outlier_map=get_outliers_by_density(means) + contribs[outlier_map]=1e-10 ## + Contribs_up.append(contribs) + Contribs_up=np.array(Contribs_up) + else: + Contribs_up=None + + print("num of gaussians: ",dicts[0]['means'].shape[0]) + # 1st Gaussian selection round + # select according to contribs + selected_ids_ts=[] + for t in range(nt): + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + else: + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) # ascending order w.r.t. ids + n=contribs.shape[0] + # candidate_ids_map=sorted_indices>=n-1000 + candidates_in_raw_ids=sorted_indices[-num_stage_wise_select_key_nodes:] #1000 + # candidates_in_raw_ids=[i for i in range(n) if candidate_ids_map[i]] + + candidate_means=means[candidates_in_raw_ids] + # candidate_contribs=contribs[candidate_ids_map] + + # voxel_size = 0.4 # 0.5 # 0.4 + voxel_grid = np.floor(candidate_means / voxel_size) + _, unique_indices = np.unique(voxel_grid, axis=0, return_index=True) + selected_candidates_in_raw_ids=[candidates_in_raw_ids[i] for i in unique_indices] + if flag_ablation_random_select: + n_random_select=len(selected_candidates_in_raw_ids) + selected_candidates_in_raw_ids=np.random.choice(selected_candidates_in_raw_ids, n_random_select, replace=False).tolist() + selected_ids_ts.append(selected_candidates_in_raw_ids) + # print(f"t={t}, selected_candidates_in_raw_ids.shape={len(selected_candidates_in_raw_ids)}") + + counts=np.zeros(n) + for t in range(nt): + selected_candidates_in_raw_ids=selected_ids_ts[t] + counts[selected_candidates_in_raw_ids]+=1 + + selected_ids_effective_1frame=[i for i in range(n) if counts[i]>0] + print("#pts left after 1st selection round: ", len(selected_ids_effective_1frame)) + # counts[selected_ids_effective_1frame] + + # avoid too many point in the 1st selection round + # select pts + def select_pts(n,selected_ids_effective_1frame): + arr1=np.ones(n) + arr2=np.zeros(len(selected_ids_effective_1frame)-n) + arr=np.concatenate((arr1,arr2),axis=0) + np.random.shuffle(arr) + selected_ids_effective_1frame=np.array(selected_ids_effective_1frame)[arr.astype(bool)].tolist() + return selected_ids_effective_1frame + # if len(selected_ids_effective_1frame)>2200: + # selected_ids_effective_1frame=select_pts(2200,selected_ids_effective_1frame) + # print("#pts left after 1st selection round (adjusted): ", len(selected_ids_effective_1frame)) + + # 2nd Gaussian selection round + ## 2nd selection round: whose contribs are effective at least nt/6 frames + # n_min_effective_frames=30 # 30 + # n_min_effective_frames=int(nt/6) + # n_min_effective_frames=min(5,int(nt/6)) # use 30 to avoid key pts not find in the long video. # TODO # 5 for haru-sit + print("n_min_effective_frames: ",n_min_effective_frames) + counts_effective_frames=np.zeros(len(selected_ids_effective_1frame)) + thresholds=[] + for t in range(nt): + if t%100==0: + print(t) + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + else: + # means=dicts[t]['means'].cpu().numpy() + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) + # candidates_in_raw_ids=sorted_indices[-2000:] + threshold=contribs[sorted_indices[-num_stage_wise_select_key_nodes]] # the 2000th largest contrib # 1000 + threshold=max(threshold,0.5) # 0.5 + counts_effective_frames+=(contribs[selected_ids_effective_1frame]>=threshold).astype(int) + thresholds.append(threshold) + + threshold_min=np.min(np.array(thresholds)) + threshold_mean=np.mean(np.array(thresholds)) + print("min contrib threshold: ",threshold_min) + print("mean contrib threshold: ",threshold_mean) + print(np.histogram(counts_effective_frames)) + if flag_show_threshold: + import matplotlib.pyplot as plt + plt.figure() + plt.plot(thresholds) + plt.show() + selected_ids_eff=[selected_ids_effective_1frame[i] for i in range(len(selected_ids_effective_1frame)) if counts_effective_frames[i]>=n_min_effective_frames] + print("#pts left after 2nd selection round: ", len(selected_ids_eff)) + # counts_effective_frames[counts_effective_frames>30].shape + + # avoid too many point in the 2nd selection round + if len(selected_ids_eff)>max_num_key_nodes: # 1000 2500 # 1500 + selected_ids_eff=select_pts(max_num_key_nodes,selected_ids_eff) # 1000 2500 1500 + print("#pts left after 2nd selection round (adjusted): ", len(selected_ids_eff)) + + if flag_ablation_select_pts_wo_uncertainty: + selected_ids_eff=np.arange(Contribs.shape[1]) + selected_ids_eff=select_pts(1000,selected_ids_eff) # 1000 # 2500 1500 + print(f"flag_ablation_select_pts_wo_uncertainty={flag_ablation_select_pts_wo_uncertainty}") + print("random select key pts:", len(selected_ids_eff)) + + if flag_show_sample: + import matplotlib.pyplot as plt + # t=nt-1 + t=0 + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + x = means[selected_ids_eff][:,0] + y = means[selected_ids_eff][:,1] + z = means[selected_ids_eff][:,2] + # sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=np.log(contribs[selected_ids_eff]),cmap='viridis') + sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=contribs[selected_ids_eff],cmap='viridis') + + cbar = plt.colorbar(sc) + cbar.set_label('Contribs') + ax.set_xlabel('X Label') + ax.set_ylabel('Y Label') + ax.set_zlabel('Z Label') + ax.axis('equal') + plt.show() + + # save + file_path=f"{folder_dict}/effective_dict_{len(selected_ids_eff)}.pkl" + effective_dict={"ids_eff2":selected_ids_eff,"ids_eff1":selected_ids_effective_1frame,"Contribs_up":Contribs_up, + "threshold_min":threshold_min,"threshold_mean":threshold_mean,"Contribs":Contribs} + file=open(file_path, 'wb') + pickle.dump(effective_dict, file) + file.close() + print(file_path) + file_effective=f"effective_dict_{len(selected_ids_eff)}.pkl" + return file_effective + +def compute_knn_indices(points, k, chunk_size=1024): + """ + Memory-efficient KNN search: returns indices of k nearest neighbors for each point. + points: [N, 3] + returns: [N, k] LongTensor + """ + N = points.shape[0] + knn_indices = [] + + for i in range(0, N, chunk_size): + end = min(i + chunk_size, N) + chunk = points[i:end] # [chunk, 3] + dists = torch.cdist(chunk, points) # [chunk, N] + topk = dists.topk(k + 1, largest=False).indices[:, 1:] # exclude self + knn_indices.append(topk) + + return torch.cat(knn_indices, dim=0) # [N, k] + +def gaussian_kde_density_knn(points, inv_covariances, k=1000, eps=1e-6): + """ + Vectorized version using k nearest neighbors for KDE density. + points: [N, 3] + covariances: [N, 3, 3] + returns: [N] density using k-NN Mahalanobis distances + """ + N, D = points.shape + device = points.device + + # Invert covariances: [N, 3, 3] + # cov_inv = torch.inverse(covariances + eps * torch.eye(D, device=device)) + cov_inv = inv_covariances + + # # Get k-NN indices using Euclidean distance + # dists = torch.cdist(points, points) # [N, N] + # knn_indices = dists.topk(k + 1, largest=False).indices[:, 1:] # [N, k] (exclude self) + + # Memory-efficient KNN + chunk_size= 10000 + knn_indices = compute_knn_indices(points, k, chunk_size=chunk_size) # [N, k] + + # Gather neighbor points: [N, k, 3] + knn_points = points[knn_indices] # batched gather + + # Center point per row: [N, 1, 3] -> broadcast to [N, k, 3] + diffs = knn_points - points[:, None, :] # [N, k, 3] + + # Apply Mahalanobis: dᵢⱼ = (xᵢ - xⱼ)^T Σᵢ⁻¹ (xᵢ - xⱼ) + # cov_inv: [N, 3, 3], diffs: [N, k, 3] + # compute: diffᵢⱼ^T @ cov_invᵢ @ diffᵢⱼ -> result [N, k] + + diffs_unsq = diffs.unsqueeze(-1) # [N, k, 3, 1] + cov_inv_exp = cov_inv[:, None, :, :] # [N, 1, 3, 3] + m = torch.matmul(cov_inv_exp, diffs_unsq) # [N, k, 3, 1] + mahalanobis_sq = torch.matmul(diffs_unsq.transpose(-1, -2), m).squeeze(-1).squeeze(-1) # [N, k] + + # Gaussian kernel weights + weights = torch.exp(-0.5 * mahalanobis_sq) # [N, k] + + density = weights.sum(dim=1) # [N] + return density + + +def plot_density_histogram_rgb(density, bins=20, height=256): + """ + density: [N] tensor + height: desired output image height in pixels + Returns: [height, width, 3] RGB uint8 numpy image of the histogram with log-scale x-axis and labeled axes + """ + density_np = density.cpu().numpy() + density_np = density_np[density_np > 0] # filter out non-positive for log scale + + dpi = 100 # Dots per inch + fig_height_inches = height / dpi + aspect_ratio = 1.5 # width / height + fig_width_inches = fig_height_inches * aspect_ratio + + # Create figure with desired size + fig, ax = plt.subplots(figsize=(fig_width_inches, fig_height_inches), dpi=dpi) + ax.hist(density_np, bins=bins, color='skyblue', edgecolor='black') + ax.set_xscale("log") + ax.set_title("Density Histogram (Log X-axis)") + ax.set_xlabel("Density (log scale)") + ax.set_ylabel("Count") + + fig.tight_layout() + fig.canvas.draw() + + width, height = fig.canvas.get_width_height() + image = np.frombuffer(fig.canvas.tostring_rgb(), dtype=np.uint8).reshape(height, width, 3) + + plt.close(fig) + return image.astype(np.uint8) + +def load_gt_masks(ws, gt_mask_dirname="mask"): + import imageio.v2 as imageio + """Load ground truth masks from directory""" + gt_mask_dir = osp.join(ws, gt_mask_dirname) + if not osp.exists(gt_mask_dir): + logging.warning(f"GT mask directory {gt_mask_dir} not found") + breakpoint() + return None + + gt_mask_fns = [ + f for f in os.listdir(gt_mask_dir) + if f.endswith(".png") or f.endswith(".jpg") + ] + gt_mask_fns.sort() + + gt_masks = [] + for gt_mask_fn in gt_mask_fns: + gt_mask_path = osp.join(gt_mask_dir, gt_mask_fn) + gt_mask = imageio.imread(gt_mask_path) + if gt_mask.ndim == 3: + gt_mask = gt_mask[..., 0] # Take first channel if RGB + gt_mask = (gt_mask > 128).astype(np.float32) # Convert to binary + gt_masks.append(gt_mask) + + gt_masks = torch.Tensor(np.stack(gt_masks)).float() # T,H,W + # self.register_gradfree_buffer("gt_masks", gt_masks) + # self.has_gt_masks = True + logging.info(f"Loaded GT masks from {gt_mask_dir}: {gt_masks.shape}, foreground ratio: {gt_masks.mean():.3f}") + return gt_masks, gt_mask_fns + +# @torch.inference_mode() +def main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model=None): + import torch + import argparse + import os + import torch + import time + from leo_renderer import Renderer + from lib_render.render_helper import render, render_gsplat + import numpy as np + from tqdm import tqdm + import imageio.v3 as iio + + from datetime import datetime + from loguru import logger as guru + import torch.nn.functional as F + import cv2 + import matplotlib.pyplot as plt + import lpips + + from lib_ugraph.ugraph_utils import compute_ssim_map, compute_lpips_map + + # parser = argparse.ArgumentParser() + # parser.add_argument("--cfg_fn", type=str) + # parser.add_argument("--port", type=int, default=8899) + # parser.add_argument("--work_dir", type=str, required=True) + # args = parser.parse_args() + + + + if dataset_name=='iphone': + cfg_fn='/data/repo/MoSca/profile/iphone/iphone_fit.yaml' + elif dataset_name=='davis' or dataset_name=='davis_mask': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='nvidia': + cfg_fn='/data/repo/MoSca/profile/nvidia/nvidia_fit.yaml' + elif dataset_name=='diffusion4d': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='objaverse': + cfg_fn='/data/repo/MoSca/profile/objaverse/objaverse_fit.yaml' + else: + cfg_fn='' + breakpoint() + port=9000 + + device = torch.device("cuda") # if torch.cuda.is_available() else "cpu") + renderer = Renderer(cfg_fn, device, work_dir, port=port, bg_flag=True, fg_flag=True,load_s2d=True, ws=ws, use_ugraph=False, relative_dir_saved_mosca_model=relative_dir_saved_mosca_model) + + nt=renderer.cams.T + print("nt:",nt) + + W, H = renderer.cams.default_W, renderer.cams.default_H + assert W == renderer.s2d.W and H == renderer.s2d.H + K = renderer.cams.K(H=H,W=W) + # focal = 0.5 * H / np.tan(0.5 * camera_state.fov).item() + # K = torch.tensor( + # [[focal, 0.0, W / 2.0], [0.0, focal, H / 2.0], [0.0, 0.0, 1.0]], + # device=device, + # ) + # w2c = torch.linalg.inv( + # torch.from_numpy(camera_state.c2w.astype(np.float32)).to(self.device) + # ) + ts=torch.arange(nt,device=device) + w2cs=torch.stack([renderer.cams.T_cw(t) for t in ts],dim=0) + # with torch.inference_mode(): + # for w2c,t in tqdm(zip(w2cs,ts)): + # # w2c = renderer.cams.T_cw(t) + + # gs5 = [] + # if renderer.bg_flag: + # gs5.append(renderer.s_model()) + # if renderer.fg_flag: + # gs5.append(renderer.d_model(t)) + # render_dict, _ = render_gsplat( + # gs5, + # H, + # W, + # K=K, + # T_cw=w2c, + # bg_color=[1.0, 1.0, 1.0], + # ) + + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # # render_dict0 = render( + # # gs5, + # # H, + # # W, + # # K=K, + # # T_cw=w2c, + # # bg_color=[1.0, 1.0, 1.0], + # # ) + + # # img0 = torch.clamp(render_dict0["rgb"].permute(1, 2, 0), 0.0, 1.0) + # # img0 = (img0.cpu().numpy() * 255.0).astype(np.uint8) + # # combined = np.hstack([img0,img]) + # # iio.imwrite(f"./test/img_{t.item()}.png", combined) # check if they look the same. + # # compare_images(img, img0) + + if save_dir=='': + save_dir = work_dir + + video_dir = f"{save_dir}/videos/{datetime.now().strftime('%Y-%m-%d-%H%M%S')}" + print("*"*20) + print("*"*20) + print(f"Saving video to {video_dir}") + print("*"*20) + print("*"*20) + os.makedirs(video_dir, exist_ok=True) + + torch.backends.cuda.preferred_linalg_library("magma") + + # if dataset_name=="davis": + # prior_depths=torch.stack(train_dataset.depths).to(device) + # elif dataset_name=="iphone": + # prior_depths=train_dataset.depths.to(device) # [nt,H,W] + # else: + # # breakpoint() + # print("No prior depth!!!") + + if dataset_name=="davis" or dataset_name=="iphone" or dataset_name=="diffusion4d" or dataset_name=="objaverse" or dataset_name=="davis_mask": + thr_density_strict=0.1 #5 + thr_density_loose=0.1 #1 + thr_rgb_diff=0.3 + thr_ssim=0.01 + elif dataset_name=="nvidia": + thr_density_strict=0.0 + thr_density_loose=0.0 + thr_rgb_diff=0.6#0.6 + thr_ssim=0.001 # 0.001 + else: + breakpoint() + + try: + # gt_mask_dir = osp.join(ws, "mask") + # renderer.s2d.load_gt_masks(gt_mask_dir) + gt_masks=renderer.s2d.gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + except: + breakpoint() + # gt_masks=None + gt_masks, gt_mask_fns=load_gt_masks(ws, "mask") + gt_masks=gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + # breakpoint() + prior_depths=renderer.s2d.dep.to(device) # [nt,H,W] + # breakpoint() + prior_depths_grad, _, _=compute_depth_gradient(prior_depths) # [nt,H,W] + # depths_info = 0.05*torch.exp(-30*prior_depths_grad) # [nt,H,W] # TODO: hyperparameter + # depths_info = depths_info.clamp_min(1e-4) # set minimum value to 1e-4 + depths_info = 1e-3 * torch.ones_like(prior_depths_grad,device=device) # 1e-3 1e-5 + # depths_info = 1e-3 * torch.ones((nt,H,W),device=device) # 1e-3 1e-5 + # depths_info[prior_depths_grad <= 0.03] = 0.1 # [nt,H,W] # TODO: hyperparameter # 0.05 # TODO!!! + print("np.histogram:",np.histogram(prior_depths_grad[gt_masks].flatten().cpu().numpy(), bins=10)) + # np.histogram(depths_info.flatten().cpu().numpy(), bins=10) + depth_min = prior_depths.min().item() + def safe_quantile(tensor: torch.Tensor, q: float, max_samples: int = 100000): + """Compute quantile safely by subsampling if needed.""" + if tensor.numel() > max_samples: + indices = torch.randperm(tensor.numel(), device=tensor.device)[:max_samples] + tensor = tensor.view(-1)[indices] + return tensor.quantile(q) + depth_max = safe_quantile(prior_depths, 0.99).item() + prior_depths_grad_threshold = safe_quantile(prior_depths_grad[gt_masks], 0.90).item() + print(f"set prior_depths_grad_threshold to {prior_depths_grad_threshold}.") + # lpips_model = lpips.LPIPS(net="alex", spatial=True).to(device) + # breakpoint() + print("fg Gaussian num:", renderer.d_model.N) + print("bg Gaussian num:", renderer.s_model.N) + print("generate out_dicts_t.pkl:") + for i, (w2c, t) in enumerate(zip(tqdm(w2cs), ts)): + assert i == t.item() + # if i<98: + # continue # test + with torch.inference_mode(): + # out_dict=renderer.model.render(int(t.item()), w2c[None], K[None], img_wh) # my code + gs5 = [] + if renderer.bg_flag: + gs5.append(renderer.s_model()) + if renderer.fg_flag: + d_list_t=renderer.d_model(t) + gs5.append(d_list_t) + n_fg = renderer.d_model(t)[0].shape[0] + render_dict, out_dict = render_gsplat( + gs5, + H, + W, + K=K, + T_cw=w2c, + bg_color=[1.0, 1.0, 1.0], + n_fg=n_fg, + ) + n_all=len(gs5) + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # density filter begin + mu, frame, scale, o, sph=d_list_t + # S = torch.zeros_like(frame) + # S[:, 0, 0] = scale[:, 0] + # S[:, 1, 1] = scale[:, 1] + # S[:, 2, 2] = scale[:, 2] + # actual_covariance = frame @ (S**2) @ frame.permute(0, 2, 1) + inv_actual_covariance = frame @ torch.diag_embed(1.0 / (scale ** 2)) @ frame.transpose(1, 2) + + # test = torch.distributions.MultivariateNormal(mu, actual_covariance).log_prob(mu) + density=gaussian_kde_density_knn(mu,inv_actual_covariance, k=100) # TODO: check scale factor + # breakpoint() + large_density_Gaussian_strict=density>thr_density_strict # TODO: threshold # 5 + large_density_Gaussian_loose=density>thr_density_loose # TODO: threshold # 1 + # density filter end + + out_dict['w2c']=w2c.detach() + out_dict['K']=K + # gt_mask = renderer.s2d.dep_mask[i].to(device) # (H, W) # test # not the true mask! + gt_mask = gt_masks[i] + # breakpoint() + # out_dict['gt_mask']=gt_mask + projected = project_points(out_dict['means'], K, w2c) # [N,2] + in_mask_Gaussians=valid_and_visible(projected, gt_mask) # [N] bool + out_dict['in_mask_Gaussians']=in_mask_Gaussians + contribs=out_dict['contribs'].reshape(-1,3)[:,0].clone() + + img = out_dict["img"][0] # [H,W,3] + gt_img = renderer.s2d.rgb[t].to(device) # [H,W,3] + diff_img = F.l1_loss(img, gt_img, reduction="none").mean(dim=-1) # [H,W] + mask_diff_img = diff_img < thr_rgb_diff # [H,W] # TODO: threshold #0.1 + small_img_diff_Gaussians=valid_and_visible(projected, mask_diff_img) # [N] bool + + # ssim map begin + ssim_map, ssim_loss=compute_ssim_map(img.cpu().numpy(),gt_img.cpu().numpy()) # [H,W,3] + mask_ssim_img = torch.tensor(np.array(ssim_map > thr_ssim),device=device).all(dim=-1) # [H,W] # TODO + large_ssim_Gaussians=valid_and_visible(projected, mask_ssim_img) # [N] bool + + # lpips_map, lpips_loss=compute_lpips_map(lpips_model, img.cpu().numpy(),gt_img.cpu().numpy(),device=device) # [H,W,1] + # mask_lpips_img = torch.tensor((lpips_map < 0.5)[:,:,0],device=device) # [H,W] # TODO + # small_lpips_Gaussians=valid_and_visible(projected, mask_lpips_img) # [N] bool + # breakpoint() + # ssim map end + # depth gradient mask + mask_depth_grad_img = prior_depths_grad[i] < prior_depths_grad_threshold # [H,W] + small_depth_grad_Gaussians=valid_and_visible(projected, mask_depth_grad_img) # [N] bool + + # + valid_gaussians_strict = small_depth_grad_Gaussians & large_ssim_Gaussians & small_img_diff_Gaussians & in_mask_Gaussians # & large_density_Gaussian_strict #& small_lpips_Gaussians # & in_mask_Gaussians # + valid_gaussians_loose = large_ssim_Gaussians & small_img_diff_Gaussians # & large_density_Gaussian_loose #& small_lpips_Gaussians # + # valid_gaussians=torch.ones(n_all, dtype=torch.bool, device=device) # TODO: threshold + # contribs_in_mask[~small_img_diff_Gaussians]=1e-8 # Gaussians on incorrect pixel are set to small contribs + contribs_in_mask=contribs.clone() + contribs_in_mask[~in_mask_Gaussians]=1e-8 # out-mask Gaussians are set to small contribs + contribs_strict=contribs.clone() + contribs_loose=contribs.clone() + contribs_strict[~valid_gaussians_strict]=1e-8 + contribs_loose[~valid_gaussians_loose]=1e-8 + + out_dict['contribs_in_mask']=contribs_in_mask + out_dict['contribs_strict']=contribs_strict + out_dict['contribs_loose']=contribs_loose + + # depth uncertainty + depths_info_perG=associate_pts_value(projected, depths_info[i]) # per Gaussian [N] + out_dict['depths_info_perG']=depths_info_perG + # out_dict['prior_depths_grad']=prior_depths_grad + + if True: + import pickle + dict_dir=f"{save_dir}/out_dicts3" + os.makedirs(dict_dir, exist_ok=True) + file=open(f'{dict_dir}/out_dict_{t}.pkl', 'wb') + pickle.dump(out_dict, file) + file.close() + + if True: + def to_rgb(img: np.ndarray) -> np.ndarray: + # [H, W] or [H, W, 1] -> [H, W, 3] + if img.ndim == 2: + return cv2.cvtColor(img, cv2.COLOR_GRAY2BGR) + elif img.ndim == 3 and img.shape[2] == 1: + return cv2.cvtColor(img[:, :, 0], cv2.COLOR_GRAY2BGR) + return img # Already RGB or unsupported format + + img_np = (img.cpu().numpy() * 255.0).astype(np.uint8) + gt_img_np = (gt_img.cpu().numpy() * 255.0).astype(np.uint8) + ssim_map_np = (np.array(ssim_map) * 255.0).clip(0, 255).astype(np.uint8) + # lpips_map_np = to_rgb((np.array(lpips_map) * 255.0).clip(0, 255).astype(np.uint8)) + + img_with_points = draw_projected_points_masked(img_np, projected, valid_gaussians_strict) # valid_gaussians, in_mask_Gaussians + img_with_points_loose = draw_projected_points_masked(img_np, projected, valid_gaussians_loose) + masked_img = overlay_mask_yellow(img_np, gt_mask.cpu().numpy()) + masked_gt_img = overlay_mask_yellow(gt_img_np, gt_mask.cpu().numpy()) + masked_diff_gt_img = overlay_mask_yellow(gt_img_np, mask_diff_img.cpu().numpy(), color= (255,0,0)) + + diff_img_np = ((diff_img*5).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # *10 to make it clear + diff_img_np = to_rgb(diff_img_np) + # depth_info_np= ((100*depths_info[t]).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # TODO: auto normalize for show + # depth_info_np = to_rgb(depth_info_np) + img_with_points_depth=draw_projected_points_colored(img_np, projected, depths_info_perG) + depth_img=apply_depth_colormap( + prior_depths[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = to_rgb(depth_img_np) + prior_depths_grad_img=apply_depth_colormap( + prior_depths_grad[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + prior_depths_grad_img_np = (prior_depths_grad_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + + # hist_img = plot_density_histogram_rgb(density,height=img_np.shape[0]) + + # combined = stack_images_side_by_side(img_with_points, masked_img) + # lpips_map_np, masked_img, masked_gt_img, hist_img , depth_info_np, depth_img_np, + combined = np.hstack([img_with_points, img_with_points_loose, depth_img_np, prior_depths_grad_img_np, ssim_map_np, masked_img, masked_gt_img, masked_diff_gt_img, diff_img_np, img_with_points_depth]) + iio.imwrite(f"{video_dir}/img_{i}.png", combined) + # breakpoint() + # del lpips_model.net + # lpips_model=None + # Delete large objects manually + del renderer + del render_dict + del inv_actual_covariance + # del lpips_model + del prior_depths + del prior_depths_grad + out_dict = {k: v.detach().cpu() if isinstance(v, torch.Tensor) else v for k, v in out_dict.items()} + del out_dict + del img, gt_img, diff_img, mask_diff_img + del projected, depths_info_perG + del contribs_strict, contribs_loose + del ssim_map, ssim_loss#, lpips_map, lpips_loss + + + # Optionally clear lists + gs5.clear() + + # Force garbage collection + import gc + gc.collect() + + # Clear PyTorch CUDA cache + torch.cuda.empty_cache() + # breakpoint() + +def compare_images(img, img0, tolerance=1): + if img.shape != img0.shape: + print(f"Shape mismatch: {img.shape} vs {img0.shape}") + return False + + diff = np.abs(img.astype(np.int32) - img0.astype(np.int32)) + max_diff = diff.max() + + if max_diff <= tolerance: + print(f"Images are close. Max diff: {max_diff}") + return True + else: + print(f"Images differ. Max diff: {max_diff}") + return False + +# def test(): +# # iphone +# # seq_name="block" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five, spin +# seq_name="backpack_" #"backpack_" +# # work_dir=f"/data/dataset/iphone_mosca_trained/paper-windmill/logs/iphone_fit_native_add3_20250226_115418" +# work_dir=f"/data/dataset/iphone_mosca_trained/backpack_/logs/iphone_fit_native_add3" #_20250223_123930" +# save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" +# dataset_name="iphone" + +# file_name_graph=f"v12_parameters8mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) +# main3(save_dir,file_effective,file_name_graph) + +# def test3(): +# # davis +# seq_name="camel" # rhino, camel, car-turn +# work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" +# save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" +# # root_dir=f"/data/dataset/custom_mocsa_initialized" +# dataset_name="davis" + +# file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 +# main3(save_dir,file_effective,file_name_graph) + + +def test2(): + # iphone + # seq_names=['paper-windmill', 'block', 'teddy', 'apple', 'haru-sit']#, 'sriracha-tree', 'mochi-high-five', ] #,'sriracha-tree','block','teddy'] + # seq_name="mochi-high-five" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five + # 'backpack_', + # seq_names=['backpack_', 'apple', 'block', 'handwavy_', 'mochi-high-five_', 'pillow_', 'spin', 'teddy', 'creeper_', 'haru-sit_', 'paper-windmill', 'space-out', 'sriracha-tree_', 'wheel'] + # seq_names=['apple', 'block', 'spin', 'teddy', 'space-out', 'wheel'] + seq_names=['apple'] + + depth_ratio=0.1 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + # auto find the folder with timestamp + # base_dir = f"/data/dataset/iphone_mosca_trained/{seq_name}/logs" + # pattern = os.path.join(base_dir, "iphone_fit_native_add3*") + # matches = glob.glob(pattern) + # if matches: + # work_dir = matches[0] # or choose with more logic + # print("Found:", work_dir) + # else: + # print("No matching folders found.") + # print('='*20) + # continue + + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/logs/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + dataset_name="iphone" + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + # main1(dataset_name,work_dir,save_dir,ws) + torch.cuda.empty_cache() + file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="max_contrib",threshold_min=0.5) #"max_contrib" minmax_distance_contrib + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + + + +def test4(): + # davis # 'car-turn', 'camel', + # seq_names=['car-turn', 'bear', 'breakdance-flare', 'dance-twirl', 'drift-straight', 'helicopter', 'libby', 'mbike-santa', 'rhino', 'schoolgirls'] # + seq_names=['camel'] + # seq_names=['car-turn', 'bear', 'drift-straight', 'helicopter', 'mbike-santa', 'rhino'] # + depth_ratio=0.001 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + dataset_name="davis" + + file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning + # main1(dataset_name,work_dir,save_dir,ws) + file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="minmax_distance_contrib",threshold_min=0.5) #"minmax_distance_contrib") + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + +def backup_code(save_dir): + from datetime import datetime + backup_dir = osp.join(save_dir, "src_backup") + os.makedirs(backup_dir, exist_ok=True) + for path in [ + "run_preprocessing.sh", + "graph_model_preprocessing5.py", + ]: + os.system(f"cp -r {path} {backup_dir}") + # reduce the backup size + # shutil.rmtree(osp.join(backup_dir, "lib_prior", "seg")) + # for root, dirs, files in os.walk(backup_dir): + # for file in files: + # if file.endswith(".pth") or file.endswith(".ckpt"): + # if osp.isfile(osp.join(root, file)): + # os.remove(osp.join(root, file)) + # else: + # shutil.rmtree(osp.join(root, file)) + # backu the commandline args + with open(osp.join(save_dir, f"commandline_args{datetime.now().strftime('%Y-%m-%d-%H%M%S')}.txt"), "w") as f: + f.write(" ".join(sys.argv)) + +def run_from_outside(): + parser = argparse.ArgumentParser() + parser.add_argument("--seq_name", type=str, required=True) + parser.add_argument("--func_name", type=str, required=True) + parser.add_argument("--dataset_name", type=str, required=True) + parser.add_argument("--depth_ratio", type=float, help="default=0.1 (iphone) or 0.001 (backpack or davis)", required=True) # default=0.1, + parser.add_argument("--threshold_min_set", type=float, help="default=0.5 or 0.1 (paper-windmill)", required=True) # default=0.5, + parser.add_argument("--version_key_edges", type=str, help="version_key_edges: max_contrib or minmax_distance_contrib", required=True) # default="max_contrib" + parser.add_argument("--extra_save_str", type=str, help="extra_save_str", required=False, default="") # _ab_uknn ## ablation study + parser.add_argument("--ratio_key", type=float, help="ratio of key nodes to determine the selected number of nodes in the first stage and threshold in the 2nd stage", required=False, default=-1) + parser.add_argument("--which_logs_folder", type=str, help="which_logs_folder", required=False, default="logs") # default="logs" + parser.add_argument("--log_subfolder", type=str, help="log_subfolder", required=False, default="demo_fit_native_add3") # default="demo_fit_native_add3" + # parser.add_argument("--use_ugraph", action="store_true") + parser.add_argument("--relative_dir_saved_mosca_model", type=str, help="relative_dir_saved_mosca_model. e.g.,", required=False) + args = parser.parse_args() + + seq_name=args.seq_name + dataset_name=args.dataset_name + depth_ratio=args.depth_ratio # 0.001 + threshold_min_set=args.threshold_min_set # 0.5 # paper-windmill 0.1 + version_key_edges=args.version_key_edges + extra_save_str=args.extra_save_str + ratio_key=args.ratio_key + # save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}/' # original + save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}{extra_save_str}/' + logs_folder=args.which_logs_folder + relative_dir_saved_mosca_model=args.relative_dir_saved_mosca_model + log_subfolder=args.log_subfolder + if dataset_name=="iphone": + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/{logs_folder}/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + voxel_size=0.2 + n_min_effective_frames=5 + # version_key_edges="max_contrib" + elif dataset_name=="davis": + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # version_key_edges="minmax_distance_contrib" + elif dataset_name=="nvidia": + work_dir=f"/data/dataset/nvidia_mosca_trained/{seq_name}/{logs_folder}/nvidia_fit_native_add3" + save_dir=f"/data/dataset/nvidia_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/nvidia_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=3 + elif dataset_name=="diffusion4d": + work_dir=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/diffusion4d_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # elif dataset_name=="objaverse": # old + # group_name="000-001" + # # /data/dataset/objaverse_mosca_trained/000-001/3ae0b4e6a75a41d38a2adf8db048a703/000/mono/logs/demo_fit_native_add3 + # work_dir=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/{logs_folder}/demo_fit_native_add3" + # save_dir=f"/data/dataset/objaverse_mosca_graph_model/{group_name}/{seq_name}/000/mono/"+save_sub_folder + # ws=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/" + # voxel_size=0.05 + # n_min_effective_frames=5 + elif dataset_name=="objaverse": + work_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/{logs_folder}/{log_subfolder}" # TODO demo_fit_native_add3_0 + save_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_graph_model/{seq_name}/mono/"+save_sub_folder + ws=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/" + voxel_size=0.1 # TODO + n_min_effective_frames=20 # TODO + elif dataset_name=="davis_mask": + work_dir=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model_masked_4x/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + else: + raise ValueError("dataset_name not supported") + + backup_code(save_dir) + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + + if args.func_name=="main1": + main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model) + elif args.func_name=="main23": + if args.depth_ratio is None or args.threshold_min_set is None or args.version_key_edges is None: + parser.error("--depth_ratio, --threshold_min_set, and --version_key_edges are required when func_name is 'main23'") + file_effective=main2(save_dir,voxel_size=voxel_size,n_min_effective_frames=n_min_effective_frames,ratio_key=ratio_key) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges=version_key_edges,threshold_min_set=threshold_min_set) #"max_contrib" minmax_distance_contrib + + + +if __name__ == "__main__": + # test2() + # test4() + # DON'T FORGET TO CHANGE! + run_from_outside() \ No newline at end of file diff --git a/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh new file mode 100644 index 0000000000000000000000000000000000000000..fade29ad45ed691be51bd485f7dd3274bb62fad1 --- /dev/null +++ b/000-004-03638bc34bae45bbbc1ca4dfe491c767-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh @@ -0,0 +1,159 @@ +#!/bin/bash + +gpu_id=3 +dataset_name=iphone + +# 'backpack_' 'haru-sit_' 'paper-windmill' +seq_names=('paper-windmill' 'block' 'teddy' 'wheel' 'apple' 'spin' 'space-out') +# seq_names=('teddy' 'wheel') +# seq_names=('spin' 'space-out') +# seq_names=('paper-windmill') +# seq_names=('block') +# seq_names=('haru-sit_' 'handwavy_' 'mochi-high-five_' 'sriracha-tree_' 'creeper_' 'pillow_') +# seq_names=('haru-sit_') # 'backpack_' + +depth_ratio=0.1 +# threshold_min_set=0.1 # 0.1 paper-windmill # 0.5 others +threshold_min_set_dict=( + ["paper-windmill"]=0.1 + ["teddy"]=0.5 + ["wheel"]=0.5 + ["apple"]=0.5 + ["spin"]=0.5 + ["space-out"]=0.5 + ["haru-sit_"]=0.5 + ["handwavy_"]=0.5 + ["mochi-high-five_"]=0.5 + ["sriracha-tree_"]=0.5 + ["creeper_"]=0.5 + ["pillow_"]=0.5 +) +version_key_edges=max_contrib # max_contrib minmax_distance_contrib max_native +ratio_key=0.02 # default -1 +extra_save_str=_ab_key_ratio_$ratio_key #2 + +# Loop and print each name +for seq_name in "${seq_names[@]}" +do + echo "${seq_name}" +done + +# Loop over each sequence +for seq_name in "${seq_names[@]}" +do + echo "Starting training for ${seq_name}" + + threshold_min_set=${threshold_min_set_dict[$seq_name]} + + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main1 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str # comment if no extra_save_str + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main23 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str \ + --ratio_key $ratio_key # ablation study + + echo "Finished training for ${seq_name}" + echo "*****************************************" + echo "*****************************************" +done + + + + + + +# gpu_id=2 +# dataset_name=davis +# seq_names=('camel' 'car-turn' 'drift-straight' 'helicopter' 'mbike-santa' 'koala' 'parkour' 'sheep' 'soccerball' 'schoolgirls') # 'bear' 'rhino' +# # seq_names=('camel' 'train' 'car-roundabout' 'car-turn' 'bike-packing' 'drift-straight' 'helicopter' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'mbike-santa' 'parkour' 'pigs' 'rhino' 'schoolgirls' 'sheep' 'soccerball' 'swing' 'bear') +# # seq_names=('train' 'car-roundabout' 'bike-packing' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'parkour' 'pigs' 'schoolgirls' 'sheep' 'soccerball' 'swing') + +# depth_ratio=0.001 # 0.1 +# threshold_min_set=0.1 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done + + + +# gpu_id=1 +# dataset_name=nvidia + +# # seq_names=('Balloon1' 'Balloon2' 'Jumping' 'Playground' 'Skating' 'Truck' 'Umbrella') # +# # seq_names=('Balloon1' 'Jumping' 'Truck' ) # 'Umbrella' +# seq_names=('Balloon2' 'Playground' 'Skating') + +# depth_ratio=0.1 +# threshold_min_set=0.01 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done \ No newline at end of file diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..f17c56412b6306cd88c4a300ef329c1a1559e666 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..c84161d8db0a3896afed9a2027a88a4d7f8ca30f Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..5c486cbb6f2c1e930c000d2caf491e9a9d96225b Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-007-02611f31791b4bc4b112cdcda8738b81-000/030/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-007-02611f31791b4bc4b112cdcda8738b81-000/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..d97c8dd1b502e3f94380f1c37f3e00aac95741f8 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-007-02611f31791b4bc4b112cdcda8738b81-000/030/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-007-02611f31791b4bc4b112cdcda8738b81-000/060/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-007-02611f31791b4bc4b112cdcda8738b81-000/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..6604dd2938546b2fb2e44525a4a5620aa8a3a4b8 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-007-02611f31791b4bc4b112cdcda8738b81-000/060/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..a7634b03bc49e4aee42eb6fefd5f6116f07414a4 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..947f8eb6de1998f2b668aecb7b0feb559ffe3cbc Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..b92ddf321245e4ac0436f8bc06458133c475d507 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..f545bd5fca83820601a1c1707261019aa1ac6173 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..b592da04204bbdb8fc1773f313f36b3aa47c1e42 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..1a6b991b69103eec7523cbb105b8da75de786d59 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..783bcc16a122e9c945b3195caae456a901eb3abc Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..b15209486e1e58f4ddb8830fa54105dbba433a34 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2141446e166b32a9184302c2b1848bfe4c20a120 Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx differ diff --git a/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..7bdf3419639963e33363111ca0871af4df7a073f Binary files /dev/null and b/000-007-02611f31791b4bc4b112cdcda8738b81-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..99548359cd288325b43851258d3f583b4d1758bd Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..0b6587e28c2a7d0217715dba5af3a69fb31696f6 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..e7c0a8af8422a84abe494477573348d8a2fb79e7 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/030/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..d701fffe79a41c289c7faa80cefdebca7502ce8a Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/030/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/060/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..a4d42c75082974551a634156ce20a15b5a6218b7 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/060/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-021458.txt b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-021458.txt new file mode 100644 index 0000000000000000000000000000000000000000..f4e0573a70a92a52443823fe213a755ec34f7c89 --- /dev/null +++ b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-021458.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-033-01dda79d3bd347bdb3326e0586aba1ff-000 --func_name main1 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-021643.txt b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-021643.txt new file mode 100644 index 0000000000000000000000000000000000000000..85c8c4a02813e9bf480c2ae41d7fdc5b46b6ea05 --- /dev/null +++ b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-021643.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-033-01dda79d3bd347bdb3326e0586aba1ff-000 --func_name main23 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..64164daf89e6d21a9fa9fae7d6e228a5c91d5a5f Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..e5a526e2eb33980e1e71ad4d38ffc1dc4bf4e1a4 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2ea2cbfec11b6f05e3985d7465edd28b574f2ab0 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..cb7c6544bf660c1187851d2c37b7c8bed74fd2cd Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..423cc36168ebd4bba6f1779b0376bae407134444 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..6d72d7bfb2b5ac9c56a649f892b3fc86f3f25280 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..cdbd45b03ad336d5594ca643ac81971ca7d13845 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..96049d930d4c4884a7bb92e573d8263db5bfebd6 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..48141498766059748e165452550a3a9b9401cde2 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..2b2e3ed0d6c92d4d70d5fc57089078a38d9cba9a Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..766bf4dc6fc351e22ade6fbbdfd19c48811e9f38 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..4cde10b9c0f80739bb861e316ec7f60a6699c26b Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt new file mode 100644 index 0000000000000000000000000000000000000000..9e1dd7bd4b614f02cae39dd8b77ea8dd499cd469 --- /dev/null +++ b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt @@ -0,0 +1,6 @@ +Function Execution Times +======================== + +photometric_reconstruct_from_pretrained: 1813.6900 seconds + +Total execution time: 1813.6900 seconds diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..1462e5759592e5c39cf3db4fe42f5bc1853e0aee Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..7b01747f74301776bfa8fd04dfb8ce948ac4b8b5 Binary files /dev/null and b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx differ diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt new file mode 100644 index 0000000000000000000000000000000000000000..3be086f34a30deb6c5a3ca7edab5455938c3e1c8 --- /dev/null +++ b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt @@ -0,0 +1 @@ +FPS: 173.1860415358 diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py new file mode 100644 index 0000000000000000000000000000000000000000..7016b5610ec6e7552d04c0af22d9b5b124beafb4 --- /dev/null +++ b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py @@ -0,0 +1,2928 @@ +import sys +import os +sys.path.append(os.path.dirname("../lib_render")) +sys.path.append(os.path.dirname("../lib_mosca/gs_utils")) +sys.path.append(os.path.dirname("..")) + +import torch +import numpy as np +from scipy.spatial.transform import Rotation +# import matplotlib.pyplot as plt +import pickle +import random +import glob + +# import pypose as pp +from torch import nn + +from itertools import chain +from scipy.spatial import KDTree,Delaunay,distance_matrix +import open3d as o3d +from collections import defaultdict +import roma + +from mpl_toolkits.mplot3d import Axes3D + +# import hdbscan +from sklearn.cluster import DBSCAN +from tqdm import tqdm + +from lib_mosca.mosca import __DQB_warp2__, __LBS_warp2__ + +from lib_mosca.gs_utils.utils_helper import project_points, draw_projected_points, draw_projected_points_masked, valid_and_visible, overlay_mask_yellow, stack_images_side_by_side, compute_depth_gradient, associate_pts_value, draw_projected_points_colored +from lib_mosca.gs_utils.som_vis_utils import ( + apply_depth_colormap, +) + +import matplotlib.pyplot as plt +import argparse +import logging, glob, sys, os, shutil, os.path as osp + +def get_order(arr): + """ + Computes the rank (order) of each element in the array. + + Args: + arr (np.ndarray): Input 1D array. + + Returns: + np.ndarray: Array where the i-th element is the rank of the i-th element in the input array. + """ + # Get the indices that would sort the array + sorted_indices = np.argsort(arr) + + # Create an array to store the ranks + order = np.empty_like(sorted_indices) + + # Assign ranks + order[sorted_indices] = np.arange(len(arr)) + + return order + +def count_out_dict_files(folder_path): + # Use glob to find files matching the pattern + file_pattern = os.path.join(folder_path, 'out_dict_*.pkl') + files = glob.glob(file_pattern) + return len(files) + +def calculate_edge_relative_orientation(Ts): # 854 pts will use ?? GB memory!!! + nk=Ts.shape[1] + Ts1=Ts.unsqueeze(2).repeat(1,1,nk,1) + Ts2=Ts.unsqueeze(1).repeat(1,nk,1,1) + Ts12=(Ts1.Inv()*Ts2) + # Ts_rel=relativelieAlgebra_pp(Ts1,Ts2) + return Ts12 + +def get_navie_ori_distances(Oris_rel): # TODO + Oris_rel_so3=Oris_rel.Log() # (nt, nk, nk, 3) + norm_oris_rel_so3=torch.norm(Oris_rel_so3.tensor(),dim=-1) # (nt, nk, nk) + norm_ori_rel_so3=torch.mean(norm_oris_rel_so3,axis=0) # (nk, nk) + return 1.0 - torch.cos(norm_ori_rel_so3) # (nk, nk) + + +def compute_accel_loss(transls): + # transls [nt,n,3] or oris [nt,n,4] + accel = 2 * transls[1:-1] - transls[:-2] - transls[2:] # [nt-2,n,3/4] + loss = accel.norm(dim=-1).mean() # [nt-2,n,3/4] => [nt-2,n]=> [1] + return loss +def cal_rigidity_loss(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global + stats = { + "loss_local_rigidity": loss_local_rigidity, + "loss_local_rigidity_ori": loss_local_rigidity_ori, + "loss_local_global": loss_local_global, + } + + return loss, stats + +def cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges): + oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + return loss_local_rigidity_dt,loss_local_rigidity_ori_dt + +def cal_rigidity_loss_extra(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + nt=transls.shape[0] + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + # 4. compute dt rigidity loss + # delta_t=50 + # oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + # oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + # transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + # transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + # dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + # error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # # 2. compute local rigidity loss for oris + # qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + # qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + # qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + # error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + if nt<100: + delta_t=int(nt/2) + else: + delta_t=50 + loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + # loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(50,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss_local_rigidity_10,loss_local_rigidity_ori_10=cal_dt_loss(10,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global+100*loss_local_rigidity_50+100*loss_local_rigidity_10#+100*loss_local_rigidity_ori_dt + stats = { + "loss_local_rigidity": loss_local_rigidity.item(), + "loss_local_rigidity_ori": loss_local_rigidity_ori.item(), + "loss_local_global": loss_local_global.item(), + "loss_local_rigidity_50": loss_local_rigidity_50.item(), + "loss_local_rigidity_10": loss_local_rigidity_10.item(), + } + + return loss, stats + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +# optimize_overall_with_orientation version of +def optimize_overall_with_orientation_edges( + edges, + Transls, Transls_optimizable, Mask, Dists_optimizable, weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, Ori_distances_optimizable, weight_ori_distance_optimizable, + Dists_observed_mean, Dists_observed_std, Mask_observed,Nearby_mask, + W2Cs,Depths_info_perG_key, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 ###### test + + # Depths_info_perG_key [nt,nk] + + nt = Transls.shape[0] + nk = Transls.shape[1] + ne = edges.shape[0] + + # Device setup + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + Dists_optimizable = nn.Parameter(Dists_optimizable.to(device)) + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Ori_distances_optimizable = nn.Parameter(Ori_distances_optimizable.to(device)) + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Mask = Mask.to(device) + Depths_info_perG_key=Depths_info_perG_key.to(device) + + optimizer = torch.optim.Adam( + [Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable, Ori_distances_optimizable], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, threshold=1e-3, min_lr=min_lr + ) + + # Distance matrix helper function + def distance_matrix(x, y): + return torch.cdist(x, y, p=2) + + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt,3,3] + # I0=torch.eye(3,device=device) + # I0[2,2]=0.1 # 0.01 camel + # I0=I0[None,:,:] + # I1=W2Cs_R.transpose(-1, -2)@I0@W2Cs_R # [nt,3,3] + # I1=I1[:,None,:,:] + I0=torch.eye(3,device=device).expand(nt,nk,-1,-1).clone() # [nt,nk,3,3] + I0[...,-1,-1]=Depths_info_perG_key # 0.01 camel # CHECK TODO + I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nk,3,3] + # breakpoint() + + # Optimization loop + # for epoch in range(max_epochs): + for epoch in tqdm(range(max_epochs), desc="Training Epochs", unit="epoch"): + optimizer.zero_grad() + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + a1=(Transls_optimizable-Transls)[:,:,None,:] # (nt, nk, 1, 3) + a1_T=a1.transpose(-1, -2) + squared_norm=(a1@I1@a1_T).squeeze(-1,-2) + # squared_norm=(a1@a1_T).squeeze(-1,-2) # test + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # add epsilon for numerical stability + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) # full model + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) # ablation of loss + # Dist_confident_pts_error_masked=Dist_confident_pts_error[Mask] + # Dist_confident_pts_error_masked_disturbed=Dist_confident_pts_error_masked+torch.randn_like(Dist_confident_pts_error_masked)*0.01 + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed) + # mask_shuffle=generate_ratio_binary_tensor(n=Dist_confident_pts_error_masked_disturbed.shape[0], ratio_0=0.8) + # mask_shuffle=torch.ones_like(Dist_confident_pts_error_masked_disturbed,dtype=torch.bool) + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed[mask_shuffle]) + + + # Ori_confident_pts_error=torch.norm(Oris_xyzw_optimizable_normed-Oris_xyzw,dim=-1) # test + # loss_oris_confident_pts = torch.mean(Ori_confident_pts_error[Mask]) + + mask_edge=generate_ratio_binary_tensor(n=edges.shape[0], ratio_0=0.3) + edges_mask=edges[mask_edge] + weight_normed=torch.abs(weight_distance_optimizable)/torch.norm(weight_distance_optimizable,dim=-1,p=1)[:,None] # (nu,nk) + weight_edges=weight_normed[edges_mask[:,0],edges_mask[:,1]] # (ne) + dis_local_canonical_edge=Dists_optimizable[edges_mask[:,0],edges_mask[:,1]] # (ne) CHECK order + # loss_rigidity,stats_rigidity=cal_rigidity_loss(Transls_optimizable,Oris_xyzw_optimizable_normed,edges,weight_edges,dis_local_canonical_edge) + loss_rigidity,stats_rigidity=cal_rigidity_loss_extra(Transls_optimizable,Oris_xyzw_optimizable_normed,edges_mask,weight_edges,dis_local_canonical_edge) + + # Loss 3: Regularization on weight distance + loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch[:-1] - Dists_epoch[1:])) # (nt-1,nk,nk) + # Dists_epoch_edges2=Dists_epoch_edges.reshape(nt,ne) # (nt,ne,nk) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch_edges2[:-1] - Dists_epoch_edges2[1:])) # (nt-1,nk,nk) + # loss_local_smoothness = torch.mean(torch.abs(Transls_optimizable[:-1]-Transls_optimizable[1:])) # (nt-1,nk,nk) # [TODO] + loss_accel = compute_accel_loss(Transls_optimizable) + + + # Loss 7: symmetry loss for edge weights + # TODO + # loss_symmetry=torch.mean(torch.abs(weight_distance_optimizable-weight_distance_optimizable.transpose(0,1))) + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + # a_loss_oris_confident_pts=1*loss_oris_confident_pts + a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point 0.1 common value + a_loss_sparsity=1e-1/loss_sparsity + a_loss_accel=0.1*loss_accel + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # flag_train_orientation=True + # if flag_train_orientation: + # a_loss_density=5e-3/loss_sparsity # 1e-3 for pos+rotation + # else: + # a_loss_density=5e-4/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + # a_loss_oris_distance_confident_pts=1*loss_oris_distance_confident_pts + # a_loss_ori_dist=1*loss_ori_dist + # a_loss_ori_local_smoothness=0.01*loss_ori_local_smoothness + + total_loss = a_loss_dist_confident_pts + a_loss_rigidity + a_loss_accel # + a_loss_sparsity + # + a_loss_dist + a_loss_local_smoothness + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness + a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # + a_loss_oris_confident_pts + + # Backpropagation and optimization step + total_loss.backward() + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # Logging progress + rigidity1=stats_rigidity["loss_local_rigidity"] + rigidity_ori2=stats_rigidity["loss_local_rigidity_ori"] + global3=stats_rigidity["loss_local_global"] + if epoch % log_interval == 0 or epoch == max_epochs - 1 or optimizer.param_groups[0]['lr'] < min_lr * 2: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + # f"a_loss_density: {a_loss_density.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + # f"OConfDist: {a_loss_oris_confident_pts.item():.6f}, " + f"a_ridigity: {a_loss_rigidity.item():.6f}, " + # f"rigidity1:{rigidity1:.6f}, " + # f"rigidity_ori2:{rigidity_ori2:.6f}, " + # f"global3:{global3:.6f}, " + + # f"Dist: {a_loss_dist.item():.6f}, " + # f"OConfDist {a_loss_oris_distance_confident_pts.item():.6f}, " + # f"ODist {a_loss_ori_dist.item():.6f}, " + # f"Smooth: {a_loss_local_smoothness.item():.6f}, " + # f"OSmooth: {a_loss_ori_local_smoothness.item():.6f}, " + # f"Rigidity L: {loss_local_rigidity_2.item():.6f}, " + f"a_loss_accel: {a_loss_accel.item():.6f}, " + f"a_loss_sparsity: {a_loss_sparsity.item():.6f}, " + # f"Symmetry L: {a_loss_symmetry.item():.6f}, " + # f"W L: {loss_weight_distance.item():.6f}, " + # f"W L2: {loss_weight_distance_2.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + print(f"weight_normed: {(weight_normed>0.1).sum().item():.6f}") + # print(stats_rigidity) + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and epoch > 5000: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + return Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable_normed, Ori_distances_optimizable + + +def get_knn_edges(n_neighbors,Transls,Contribs,Transls2,Contribs2,contrib_threshold, flag_remove_self=False): + # choice 2: knn graph + def o3d_knn_p(pts2,pts, num_knn, flag_remove_self=False): + indices = [] + sq_dists = [] + pcd = o3d.geometry.PointCloud() + pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + pcd_tree = o3d.geometry.KDTreeFlann(pcd) + for p in pts2: + if not flag_remove_self: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + indices.append(i[0:]) + sq_dists.append(d[0:]) + else: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn + 1) + indices.append(i[1:]) + sq_dists.append(d[1:]) + return np.array(sq_dists), np.array(indices) + + nk=Transls.shape[1] + nu=Transls2.shape[1] + + i_s_knn_others,i_s_knn_is=[],[] + # n_neighbors=5 + for i in range(nu): + t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + pts2=[pt2] # just one point + sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors, flag_remove_self) + sq_dists,indices= sq_dists[0],indices[0] + i_s_knn=i_s[indices] # usage: ids[i_s_knn] + i_s_knn_others.append(i_s_knn) + i_s_knn_is.append(i*torch.ones_like(i_s_knn)) + i_s_knn_is=torch.cat(i_s_knn_is) + i_s_knn_others=torch.cat(i_s_knn_others) + edges_others=torch.stack([i_s_knn_is,i_s_knn_others],dim=-1) + # breakpoint() + # old version: has bug when knn key points are not enough + # i_s_knn_others=torch.stack(i_s_knn_others) + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # edges_others.shape # nu*5 + return edges_others + +# def get_knn_edges_naive(n_neighbors,Transls,Transls2, flag_remove_self=False): +# nk=Transls.shape[1] +# nu=Transls2.shape[1] +# nt=Transls.shape[0] +# Dist=[] +# for t in range(nt): +# dist=(Transls2[t][:,None]-Transls[t][None]).norm(dim=-1) +# Dist.append(dist) +# Dist=torch.stack(Dist,dim=0) +# Dist=Dist.mean(dim=0) + +# def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): +# """ +# Naive temporal k-NN graph construction using average distance over time. + +# Args: +# n_neighbors (int): Number of neighbors. +# Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. +# Transls2 (torch.Tensor): (T, N2, D) tensor of query points. +# flag_remove_self (bool): If True and N1 == N2, removes self-matches. + +# Returns: +# torch.Tensor: (N2 * n_neighbors, 2) edge index pairs [Transls2_index, Transls_index]. +# """ +# T, N1, D = Transls.shape +# N2 = Transls2.shape[1] + +# # Compute temporal average pairwise distances +# dist_list = [] +# for t in range(T): +# dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) +# dist_list.append(dist) +# Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + +# if flag_remove_self and N1 == N2: +# diag = torch.arange(N2, device=Dist.device) +# Dist[diag, diag] = float('inf') + +# # Get top-k nearest neighbors for each Transls2 point +# _, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + +# # Build edge index pairs +# row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) +# edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + +# return edges_others + +def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): + """ + Naive temporal k-NN graph construction using average distance over time. + + Args: + n_neighbors (int): Number of neighbors. + Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. + Transls2 (torch.Tensor): (T, N2, D) tensor of query points. + flag_remove_self (bool): If True and N1 == N2, removes self-matches. + + Returns: + edges_others (torch.Tensor): (N2 * k, 2) edge index pairs [Transls2_index, Transls_index]. + dists (torch.Tensor): (N2 * k,) mean distances corresponding to each edge. + """ + T, N1, D = Transls.shape + N2 = Transls2.shape[1] + + # Compute temporal average pairwise distances + dist_list = [] + for t in range(T): + dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) + dist_list.append(dist) + Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + + if flag_remove_self and N1 == N2: + diag = torch.arange(N2, device=Dist.device) + Dist[diag, diag] = float('inf') + + # Get top-k nearest neighbors and distances + dists, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + + # Build edge index pairs + row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) + edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + dists_edges = dists.reshape(-1) # (N2 * k,) + + return edges_others, dists_edges + + + + + +def optimize_others_with_orientation_edges4( + edges, + Transls, Transls_optimizable, Mask, #Dists_optimizable, #weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, #Ori_distances_optimizable, #weight_ori_distance_optimizable, + Transls_key, Oris_xyzw_key,Transls_key_optimized, Oris_xyzw_key_optimized, + weight_param_key_optimizable,sq_dists_key_nonkey_edges, + ts_most_confident,per_nonkeyGaussian_dweight, + W2Cs,Depths_info_perG_nonkey, + device, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 + # weight_param_key_optimizable [nk] per-keyGuassian weight + # Depths_info_perG_nonkey [nt,nu] + # nt = ts.shape[0] + + + nt_all = Transls.shape[0] + nu=Transls.shape[1] + nk=Transls_key_optimized.shape[1] + ne=edges.shape[0] + + Oris_xyzw=Oris_xyzw/Oris_xyzw.norm(dim=-1,keepdim=True) # normalize + # Oris_xyzw_optimizable=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) # normalize + Oris_xyzw_key_optimized=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) # normalize + + # Device setup + # Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + # Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Transls_optimizable=Transls_optimizable.to(device) + Oris_xyzw_optimizable=Oris_xyzw_optimizable.tensor().to(device) + + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Transls_key = Transls_key.to(device) + Oris_xyzw_key = Oris_xyzw_key.to(device) + Transls_key_optimized = Transls_key_optimized.to(device).detach() + Oris_xyzw_key_optimized = Oris_xyzw_key_optimized.to(device).detach() + weight_param_key_optimizable = nn.Parameter(weight_param_key_optimizable.to(device)) + # sq_dists_key_nonkey = sq_dists_key_nonkey.to(device) + sq_dists_key_nonkey_edges = sq_dists_key_nonkey_edges.to(device) + per_nonkeyGaussian_dweight = nn.Parameter(per_nonkeyGaussian_dweight.to(device)) # don't use for now, just keep + # per_nonkeyGaussian_dweight=None + Depths_info_perG_nonkey = Depths_info_perG_nonkey.to(device) + + # sq_dists_key_nonkey_edges=sq_dists_key_nonkey[edges[:,0],edges[:,1]] # (ne) + exp_sq_dists_key_nonkey_edges=torch.exp(-sq_dists_key_nonkey_edges) # (ne) + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt_all,3,3] + + Mask = Mask.to(device) + + # Distance matrix helper function + # def distance_matrix(x, y): + # return torch.cdist(x, y, p=2) + + @torch.no_grad() # check + def cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu): + # weight_param_key_optimizable [nk] + # exp_sq_dists_key_nonkey_edges [ne] + device=weight_param_key_optimizable.device + nk=weight_param_key_optimizable.shape[0] + + weight_param_key_optimizable_k=100*torch.abs(weight_param_key_optimizable) # [TODO]: 2000 # 100 + weight_param_edges=weight_param_key_optimizable_k[edges[:,1]] # (ne) + # assert all(weight_param_edges>0) + # assert all(sq_dists_key_nonkey_edges>0) + # weight_edges = (exp_sq_dists_key_nonkey_edges**(1/10))**weight_param_edges # TODO: to unify scale #test: 10 for block + weight_edges = exp_sq_dists_key_nonkey_edges**weight_param_edges # TODO: to unify scale #test: 20 for block + + weight_edges = torch.clamp(weight_edges, min=1e-10, max=1e10) # test + + weights_matrix=torch.zeros((nu,nk),device=device) + weights_matrix[edges[:,0],edges[:,1]]=weight_edges + sum_temp = torch.sum(weights_matrix,dim=-1,keepdim=True) + if any(sum_temp==0): + breakpoint() + if torch.isnan(weight_param_key_optimizable).any(): + breakpoint() + weights_matrix_normed=weights_matrix/sum_temp + + weight_edges_normed=weights_matrix_normed[edges[:,0],edges[:,1]] # (ne) + + return weight_edges_normed + + + + # prepare + Oris_xyzw_key_norm=Oris_xyzw_key/Oris_xyzw_key.norm(dim=-1,keepdim=True) + Oris_xyzw_key_optimized_norm=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) + n_neighbors=(edges[:,0]==0).sum() + if n_neighbors!=5: + breakpoint() + + # new + src_xyz_1=Transls[ts_most_confident,edges[::n_neighbors,0]] # (nu,3) + src_xyzw_1=Oris_xyzw[ts_most_confident,edges[::n_neighbors,0]] # (nu,4) + src_xyz_1 = nn.Parameter(src_xyz_1) + src_xyzw_1 = nn.Parameter(src_xyzw_1) + + # get sk_src_node and sk_dst_node (key nodes) + ts_most_confident_repeat=ts_most_confident.repeat(n_neighbors,1).T.reshape(-1) # (nt*5) + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + # optimizer = torch.optim.Adam( + # [weight_param_key_optimizable,src_xyz_1,src_xyzw_1,per_nonkeyGaussian_dweight], lr=lr + # ) + optimizer = torch.optim.Adam( + [weight_param_key_optimizable,src_xyz_1,src_xyzw_1], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, min_lr=min_lr + ) + ######################################################3 + + # ts_batches = torch.split(ts_all, batch_size) + # for ts_batch in ts_batches: + # Optimization loop + # for epoch in range(max_epochs): + # ts_all=torch.arange(nt_all,device=device) + # batch_size=30 + # pbar=tqdm(range(max_epochs)) + # for step in pbar: + # optimizer.zero_grad() + ts_all = torch.arange(nt_all, device=device) + batch_size = 30 + steps_per_epoch = (nt_all + batch_size - 1) // batch_size + total_steps = steps_per_epoch * max_epochs + + pbar = tqdm(range(total_steps), desc="Training") + + for step in pbar: + epoch = step // steps_per_epoch + inner_step = step % steps_per_epoch + + # Shuffle once per epoch + if inner_step == 0: + ts_shuffled = ts_all[torch.randperm(nt_all)] + + # Select current batch + start = inner_step * batch_size + ts_batch = ts_shuffled[start:start + batch_size] + pbar.set_description(f"Epoch {epoch}, Step {inner_step}/{steps_per_epoch-1}") + + optimizer.zero_grad() + + # print(f"ts_batch: {ts_batch}") + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # Mask_DQB=Mask[ts_batch] + + sk_nt_node_xyz_1=Transls_key[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_2=sk_nt_node_xyz_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_1=Oris_xyzw_key_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_1=sk_nt_node_xyzw_1[:,:,[3,0,1,2]] + sk_nt_node_quat_2=sk_nt_node_wxyz_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,4) + sk_nt_dst_node_xyz=sk_nt_node_xyz_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat=sk_nt_node_quat_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # error propagation + I0=torch.eye(3,device=device).expand(nt,nu,-1,-1).clone() # [nt,nu,3,3] + # I0[...,-1,-1]=Depths_info_perG_nonkey # 0.001 camel # CHECK TODO + # I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nu,3,3] + I1=I0 # test # [nt,nu,3,3] + I1_masked=I1[Mask_DQB] # [n_mask,3,3] + + if torch.isnan(weight_param_key_optimizable).any(): + print("weight_param_key_optimizable is nan") + breakpoint() + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + src_xyz=src_xyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask<=nt*nu,4) + + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) # [nu,n_neighbors] + # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # sk_w=weight_per_nonkeyGaussian_adjusted.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz, # [N,K,3] + sk_nt_dst_node_quat, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + # Transls_interpolation=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + # Oris_interpolation=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_interpolation_masked=mu_dst # [n_mask,3] + Oris_interpolation_masked=quat_dst_xyzw # [n_mask,4] + + + # error propagation + # a1=(Transls_interpolation[Mask]-Transls[Mask])[:,None,:] # (n_mask, 1, 3) + a1=(Transls_interpolation_masked-Transls[ts_batch][Mask_DQB])[:,None,:] # (n_mask, 1, 3) + # a1=(Transls_interpolation_masked-Transls[ts_batch])[:,None,:] # (n_mask, 1, 3) # ablation of loss + a1_T=a1.transpose(-1, -2) # (n_mask, 3, 1) + squared_norm=(a1@I1_masked@a1_T).squeeze(-1,-2) # [n_mask] + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # [n_mask] # add epsilon for numerical stability + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) # [nt,nu] + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) + # Dist_confident_pts_error=torch.norm(Transls_interpolation[Mask]-Transls[Mask],dim=-1) # [nt,nu] # last one + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Ori_confident_pts_error=torch.norm(Oris_interpolation[Mask]-Oris_xyzw[Mask],dim=-1) # [n_mask] + Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] + # Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] # ablation of loss + loss_ori_confident_pts = torch.mean(Ori_confident_pts_error) + + # Loss 3: Regularization on weight distance + # loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_velocity = torch.mean(torch.abs(Transls_interpolation[:-1]-Transls_interpolation[1:])) # (nt-1,nk,nk) # [TODO] + # loss_ori_velocity = torch.mean(torch.abs(Oris_interpolation[:-1] - Oris_interpolation[1:])) + + # loss_accel = compute_accel_loss(Transls_interpolation) + # loss_ori_accel=compute_accel_loss(Oris_interpolation) + + + # Loss 6: regularize per nonkey Gaussian delta weight + # loss_per_nonkeyGaussian=torch.mean(torch.abs(per_nonkeyGaussian_dweight)) + # loss_per_nonkeyGaussian = 1000*torch.mean(per_nonkeyGaussian_dweight ** 2) # TODO + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + a_loss_ori_confident_pts=1*loss_ori_confident_pts + # a_loss_interpolation=1*loss_interpolation # 100 + # a_loss_interpolation_ori=1*loss_interpolation_ori # 100 + # a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point + + # a_loss_velocity=0.1*loss_velocity + # a_loss_ori_velocity=0.1*loss_ori_velocity + # a_loss_accel=0.1*loss_accel + # a_loss_ori_accel=0.1*loss_ori_accel + + # a_loss_per_nonkeyGaussian=1e-2*loss_per_nonkeyGaussian + + # a_loss_sparsity=1e-3*loss_sparsity + + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + + total_loss = a_loss_dist_confident_pts + a_loss_ori_confident_pts #+ a_loss_per_nonkeyGaussian#+ \ + # a_loss_velocity + a_loss_ori_velocity + a_loss_accel + a_loss_ori_accel #+ a_loss_sparsity \ + # a_loss_dist + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness+ a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # a_loss_rigidity + # a_loss_interpolation + a_loss_interpolation_ori + \ + + # Backpropagation and optimization step + total_loss.backward() + # total_loss.backward(retain_graph=True) + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # # Logging progress + if epoch % log_interval == 0 or epoch == max_epochs - 1: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + f"a_loss_ori_confident_pts: {a_loss_ori_confident_pts.item():.6f}, " + # f"a_loss_per_nonkeyGaussian: {a_loss_per_nonkeyGaussian.item():.6f}, " + # f"a_loss_interpolation: {a_loss_interpolation.item():.6f}, " + # f"a_loss_interpolation_ori: {a_loss_interpolation_ori.item():.6f}, " + # f"a_loss_rigidity: {a_loss_rigidity.item():.6f}, " + # f"a_loss_velocity: {a_loss_velocity.item():.6f}, " + # f"a_loss_ori_velocity: {a_loss_ori_velocity.item():.6f}, " + # f"a_loss_accel: {a_loss_accel.item():.6f}, " + # f"a_loss_ori_accel: {a_loss_ori_accel.item():.6f}, " + # f"Density L: {a_loss_density.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + # # print(f"weight_normed: {(weight_normed>1e-6).sum().item():.6f}") + # # print(f"weight_param_key_optimizable:" {weight_param_key_optimizable}) + + pbar.set_description(f"L:{total_loss.item():.6f}, "+ + f"c:{a_loss_dist_confident_pts.item():.6f}, "+ + f"oc:{a_loss_ori_confident_pts.item():.6f}, "+ + # f"per:{a_loss_per_nonkeyGaussian.item():.6f}, "+ + # f"i: {a_loss_interpolation.item():.6f}, "+ + # f"io: {a_loss_interpolation_ori.item():.6f}, "+ + # f"r:{a_loss_rigidity.item():.6f}, "+ + # f"b:{a_loss_velocity.item():.6f}, "+ + # f"ov:{a_loss_ori_velocity.item():.6f}, "+ + # f"a:{a_loss_accel.item():.6f}, "+ + # f"oa:{a_loss_ori_accel.item():.6f}, "+ + f"lr:{optimizer.param_groups[0]['lr']:.2e}") + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + # # TEMP: for keeping data structure + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + # sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) + # # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # # sk_w=weight_per_nonkeyGaussian_adjusted[None,:,:].repeat(nt,1,1)[Mask_DQB] # (n_mask,n_neighbors) + + # weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + # weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + # # for i,e in enumerate(edges): + # # weight_distance_optimizable[e[0],e[1]]=weight_edges[i] + # weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + # print("End of optimize_others_with_orientation_edges4.") + # # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + # print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + # blend_dict = { + # 'sk_w': sk_w, + # 'src_xyz': src_xyz, + # 'src_quat': src_quat, + # 'sk_nt_src_node_xyz': sk_nt_src_node_xyz, + # 'sk_nt_src_node_quat': sk_nt_src_node_quat, + # 'sk_nt_dst_node_xyz': sk_nt_dst_node_xyz, + # 'sk_nt_dst_node_quat': sk_nt_dst_node_quat + # } + + + + # ********************* new trying begin ******************* + # Step 1: calculate src_xyz_1_new and src_xyzw_1_new + weight_param_key_optimizable=weight_param_key_optimizable.abs() + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # (nu,n_neighbors) + + weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + + # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + sk_w=weight_per_nonkeyGaussian # (nu,n_neighbors) + + sk_dst_node_xyz_1=Transls_key_optimized[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_dst_node_xyzw_1=Oris_xyzw_key_optimized_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_dst_node_wxyz_1=sk_dst_node_xyzw_1[:,[3,0,1,2]] + sk_dst_node_xyz_2=sk_dst_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_dst_node_quat_2=sk_dst_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + # breakpoint() + mu_dst, quat_dst=__DQB_warp2__( # [nu,3], [nu,4] + sk_w, # [N,K] + src_xyz_1, # [N,3] + sk_src_node_xyz_2, # [nu,n_neighbors,3] + sk_src_node_quat_2, # [nu,n_neighbors,4] + sk_dst_node_xyz_2, # [N,K,3] + sk_dst_node_quat_2, # [N,K,4] + dyn_o=None, + src_quat=src_wxyz_1, # [N,4] # wxyz + ) + src_xyz_1_new=mu_dst + src_xyzw_1_new=quat_dst[:,[1,2,3,0]] # [nu,4] + # breakpoint() + # for further optimization + blend_opt_dict = { + 'weight_param_key_optimizable': weight_param_key_optimizable, # (nu,n_neighbors) + 'src_xyz_1': src_xyz_1_new, # (nu,3) # CHECK + 'src_xyzw_1': src_xyzw_1_new, # (nu,4) # CHECK + 'per_nonkeyGaussian_dweight': per_nonkeyGaussian_dweight, # (nu,n_neighbors) + 'exp_sq_dists_key_nonkey_edges': exp_sq_dists_key_nonkey_edges, # (ne) + 'ts_most_confident': ts_most_confident, # (nu) + } + + # Step 2: calculate Transls_optimized and Oris_xyzw_optimized + Transls_optimized,Oris_xyzw_optimized=[],[] + ts_batches = torch.split(ts_all, batch_size) + for ts_batch in ts_batches: + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,4) + + # key points dst needed to be updated; src is the same + sk_nt_node_xyz_op_1=Transls_key_optimized[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_op_2=sk_nt_node_xyz_op_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_op_1=Oris_xyzw_key_optimized_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_op_1=sk_nt_node_xyzw_op_1[:,:,[3,0,1,2]] + sk_nt_node_quat_op_2=sk_nt_node_wxyz_op_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_dst_node_xyz_op=sk_nt_node_xyz_op_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat_op=sk_nt_node_quat_op_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # nonkey points update when src_xyz_1 and src_xyzw_1 are trainable + src_xyz=src_xyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask<=nt*nu,4) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz_op, # [N,K,3] + sk_nt_dst_node_quat_op, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + Transls_optimized_batch=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + Oris_xyzw_optimized_batch=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_optimized.append(Transls_optimized_batch) + Oris_xyzw_optimized.append(Oris_xyzw_optimized_batch) + Transls_optimized=torch.cat(Transls_optimized,dim=0) # [nt_all,nu,3] + Oris_xyzw_optimized=torch.cat(Oris_xyzw_optimized,dim=0) # [nt_all,nu,4] + + # ********************* new trying end ******************* + # breakpoint() + blend_dict={} + return Transls_optimized, Oris_xyzw_optimized, weight_param_key_optimizable,weight_distance_optimizable,blend_dict,blend_opt_dict + +def get_knn_edges_key(Transls,Mask,k): + # group by variance of relative motion + # Transls [nt,nk,3] + # std_threshold=0.01 + # max_dist_threshold=0.4 + + def nanstd(o, dim, keepdim=False): + result = torch.sqrt( + torch.nanmean( + torch.pow( torch.abs(o-torch.nanmean(o,dim=dim).unsqueeze(dim)),2), + dim=dim + ) + ) + if keepdim: + result = result.unsqueeze(dim) + return result + + nu=Transls.shape[1] + + # Dist=torch.cdist(Transls, Transls, p=2) + Dist = torch.norm(Transls[:, :, None, :] - Transls[:, None, :, :], dim=-1) # [nt, nu, nk] + Mask_dist=Mask[:, :, None] * Mask[:, None, :] # [nt, nu, nk] + Dist_masked = torch.where(Mask_dist, Dist, torch.tensor(float('nan'), device=Dist.device)) + # std_dist=Dist.std(dim=0) # [nu,nk] + std_dist=nanstd(Dist_masked,dim=0) # [nu,nk] + + mean_dist_masked=Dist_masked.nanmean(dim=0) # [nu,nk] # may include nan + # mean_dist=Dist.mean(dim=0) # [nu,nk] + Dist_masked2 = torch.where(Mask.unsqueeze(1).expand(-1, nu, -1), Dist, torch.tensor(float('nan'), device=Dist.device)) + mean_dist_masked2=Dist_masked2.nanmean(dim=0) # [nu,nk] # should not include nan # may still have nan + # mean_dist_masked2=Dist.mean(dim=0) # [nu,nk] # overwrite [TEST] + # breakpoint() + assert torch.any(mean_dist_masked2.isnan())==False + mean_dist_masked = torch.where(~torch.isnan(std_dist), mean_dist_masked, mean_dist_masked2) + # std_dist[~torch.isnan(std_dist)].mean(),mean_dist[~torch.isnan(mean_dist)].mean() + Dist_masked2 = torch.where(Mask_dist, Dist_masked, torch.tensor(float('-inf'), device=std_dist.device)) + # np.histogram(std_dist.numpy(),bins=50) + # np.histogram(std_dist[~torch.isnan(std_dist)].cpu().numpy(), bins=50) + # std_dist.min() + + max_dist = torch.max(Dist_masked2, dim=0).values # has -inf + max_dist2 = torch.where(~torch.isinf(max_dist), max_dist, torch.tensor(float('inf'), device=max_dist.device)) + # np.histogram(max_dist2.numpy()) + # np.histogram(max_dist[~torch.isinf(max_dist)].cpu().numpy(), bins=10) + + # (std_dist<0.01).sum(),(max_dist[~torch.isinf(max_dist)]<0.4).sum(),(max_dist2<0.4).sum(),torch.isnan(std_dist).sum() + + # try 1: + # connectivity_matrix=(std_dist40: + conditiona=Contribs2>0.5*np.max(Contribs2,axis=0) # 0.707 # TODO + # conditionb=Contribs2>contrib_threshold + # condition2=np.logical_or(conditiona,conditionb) + condition2=conditiona + else: + conditionb=Contribs2>contrib_threshold + condition2=conditionb + Mask2=torch.tensor(condition2,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable2=Oris_xyzw2.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edges + # 2.1 position edges + Dists_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True) + + # 2.2 orientation edges + Ori_distances_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True)*2 + + # 3. edge weight + # 3.1 position stds + weight_distance_optimizable2=torch.zeros((nu,nk),dtype=torch.float32,requires_grad=True) + + # knn graph (max contribs version) + + # def o3d_knn_p(pts2,pts, num_knn): + # indices = [] + # sq_dists = [] + # pcd = o3d.geometry.PointCloud() + # pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + # pcd_tree = o3d.geometry.KDTreeFlann(pcd) + # for p in pts2: + # [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + # indices.append(i[0:]) + # sq_dists.append(d[0:]) + # return np.array(sq_dists), np.array(indices) + + + # i_s_knn_others=[] + # sq_dists_5=[] + # n_neighbors=5 # 5 # 10 can not work + # ts_most_confident=[] + # for i in range(nu): + # t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + # eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + # pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + # i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + # pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + # pts2=[pt2] # just one point + # sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors) + # sq_dists,indices= sq_dists[0],indices[0] + + # flag_use_key_point_key_neighbors=True + # if flag_use_key_point_key_neighbors: + # ## given a key point, get the knn n_neighbors key points + # indice_nearest_key=indices[0] # nearest key point index + # pt3=pts[indice_nearest_key] # nearest key point + # pts3=[pt3] # just one key point + # sq_dists, indices = o3d_knn_p(pts3, pts, n_neighbors) # inlcude itself + # sq_dists,indices= sq_dists[0],indices[0] + # ## + # # calcuate the distance between the i-th other point and the neighboring key points + # sq_dists=np.linalg.norm(pts[indices]-pt2,axis=-1) # [n_neighbors] + + + # i_s_knn=i_s[indices] # usage: ids[i_s_knn] + # i_s_knn_others.append(i_s_knn) + # sq_dists_5.append(sq_dists) + # ts_most_confident.append(t_most_confident) + # i_s_knn_others=torch.stack(i_s_knn_others) + # sq_dists_5=np.array(sq_dists_5) + # sq_dists_5=torch.tensor(sq_dists_5,dtype=torch.float32) + # ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # # edges_others.shape # nu*5 + + # knn graph (minmax distance + contrib version) + n_neighbors=5 + edges_others,mean_dists=get_knn_edges_others(Transls,Mask,Transls2,Mask2,k=n_neighbors) + sq_dists_5=mean_dists[edges_others[:,0],edges_others[:,1]].reshape(-1,n_neighbors) # [nu,n_neighbors] + # edges_others,dists_edges=get_knn_edges_naive(n_neighbors,Transls,Transls2,flag_remove_self=False) # ablation of uknn + # sq_dists_5=dists_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + ts_most_confident=np.argmax(Contribs2[:,:],axis=0) # [nu] + ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + torch.cuda.empty_cache() + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + + # Method 3: + # use Dual Quat Blending (refer to MoSca) + # sq_dists_key_nonkey=-torch.ones([nu,nk]) + # sq_dists_key_nonkey[edges_others[:,0],edges_others[:,1]]=sq_dists_5.reshape(-1) + sq_dists_key_nonkey_edges=sq_dists_5.reshape(-1) # [ne] + weight_param_key_optimizable=1*torch.ones(nk,dtype=torch.float32,requires_grad=True) # 2000 [TODO] + per_nonkeyGaussian_dweight=torch.zeros(nu,n_neighbors,dtype=torch.float32,requires_grad=True) + + # batch training along t + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + + Transls_optimized2,Oris_xyzw_optimized2, weight_param_key_optimized2,weight_distance_optimized2,blend_dict,blend_opt_dict=\ + optimize_others_with_orientation_edges4(\ + edges_others,\ + Transls2, Transls_optimizable2.clone(), Mask2, \ + Oris_xyzw2, Oris_xyzw_optimizable2.clone(), \ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + weight_param_key_optimizable,sq_dists_key_nonkey_edges,\ + ts_most_confident,per_nonkeyGaussian_dweight,\ + W2Cs,Depths_info_perG_nonkey, + device) + + + # keep + edges_others_optimized=edges_others + Dists_optimized2=Dists_optimizable2 + Ori_distances_optimized2=Ori_distances_optimizable2 + + return Transls_optimized2,Dists_optimized2,weight_distance_optimized2,Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others,weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict + + + + + + +def main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges,threshold_min_set): + import sys + import os + sys.path.append(os.path.dirname("../flow3d")) + + import torch + import numpy as np + from scipy.spatial.transform import Rotation + import matplotlib.pyplot as plt + import pickle + import random + import glob + + import pypose as pp + from torch import nn + + from itertools import chain + from scipy.spatial import KDTree,Delaunay,distance_matrix + import open3d as o3d + from collections import defaultdict + import roma + + # set args + + # work_dir="/data/dataset/used_ds/som/sriracha-tree/work-dir" # backpack,haru-sit, sriracha-tree + # work_dir="/data/dataset/used_ds/som/backpack/work-dir" # work-dir1 # initialization results (no SoM model) + # work_dir="/data/dataset/custom_ds_trained/rhino" + # work_dir="/data/dataset/custom_ds_graph_model/rhino" + work_dir=save_dir + + # file_effective="effective_dict.pkl" # backpack 854 + # file_effective="effective_dict_949.pkl" # haru-sit + # file_effective="effective_dict_1908.pkl" # sriracha-tree # 1908 # 1000 + # file_effective="effective_dict_484.pkl" # backpack # work-dir1 + # file_effective="effective_dict_712.pkl" # rhino + + # dict_dir=work_dir+"/out_dicts2" # has w2c and camera's K + dict_dir=work_dir+"/out_dicts3" # has w2c and camera's K + + flag_draw_contribs_sample=False + flag_visulize_sparisity=False + flag_visulize_key_point=False + flag_visulize_nonkey_point=False + + flag_auto_threshold=False + contrib_threshold=2 # usually 2 + threshold=2e-0 ## TODO # for non-key points # fixed threshold + # threshold=-torch.inf + + + + # [1] load data + nt=count_out_dict_files(dict_dir) + dicts=[] + for t in range(nt): + with open(f"{dict_dir}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # print(dicts[0].keys()) + + with open(f"{dict_dir}/{file_effective}", 'rb') as file: + effective_dict = pickle.load(file) + inds_effective = effective_dict['ids_eff2'] #ids_eff1 + # Contribs_up = effective_dict['Contribs_up'] + threshold_min = effective_dict['threshold_min'] + threshold_mean = effective_dict['threshold_mean'] + Contribs_all = effective_dict['Contribs'] + print(f"threshold_min:{threshold_min}, threshold_mean:{threshold_mean}") + + # threshold_min=np.max([threshold_min-1, 0.5]) # 0.5 + # threshold_min=np.max([threshold_min-3, 0.5]) # 0.5 # paper-windmill test + # threshold_min=0.1 # test: believe nearly all points # paper-windmill + # threshold_min=0.5 # test: camel,backpack + if threshold_min_set!=-1: + print(f"set threshold_min to {threshold_min_set}.") + threshold_min=threshold_min_set + else: + print(f"use threshold_min from effective_dict: {threshold_min}.") + + # breakpoint() + if flag_auto_threshold: + print(f"set auto threshold to {threshold_min}.") + contrib_threshold=threshold_min + threshold=threshold_min + # contrib_threshold=threshold_mean + # threshold=threshold_mean + + print(f"set threshold_min to {threshold_min}.") + + device='cuda' + ids=inds_effective + # ids=np.array(inds_effective)[np.arange(0,200,6)].tolist() # quick test + # ids=np.array(inds_effective)[np.arange(0,854,2)].tolist() + + i_target=0 + i_show=10 + Transls=np.array([dicts[t]["means"][ids].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz=np.array([dicts[t]["quats"][ids].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw=Quats_wxyz[...,[1,2,3,0]] # xyzw + # Contribs=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) #### + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids] for t in range(nt)]) #### + Contribs=Contribs_all[:,ids] + # Contribs=Contribs_up[:,ids] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort=np.array([get_order(Contribs[:,i]) for i in range(Contribs.shape[1])]).transpose(1,0) + # Vars=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) + W2Cs=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + SE3s=np.concatenate([Transls,Quats_xyzw],axis=-1) # xyz xyzw (required by pypose) + SE3s=torch.tensor(SE3s,dtype=torch.float32) #,device='cuda') + Oris_xyzw=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type) + Oris=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type).Log() + lie_tensor_SE3s=pp.LieTensor(SE3s, ltype=pp.SE3_type) + Depths_info_perG=torch.stack([dicts[t]['depths_info_perG'].cpu() for t in range(nt)]) + # depth_ratio=0.001 # 0.001 + Depths_info_perG=depth_ratio*torch.ones_like(Depths_info_perG) # TEST + print(f"TEST: set Depths_info_perG={depth_ratio}*torch.ones_like(Depths_info_perG).") # test + Depths_info_perG_key=Depths_info_perG[:,ids] + # len(dicts) + # vars=Vars[:,i_target] + nk=len(ids) # num of key points + # np.histogram(vars[vars<1e0]) + print("nk:",nk) + print("dicts[0]['means'].shape:", dicts[0]["means"].shape) + + # draw + if flag_draw_contribs_sample: + # i_show=20 + # means=Transls[:,i_show] + # vars0=Vars[:,i_show] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars0,axis=1)),cmap='viridis') + + # means=Transls[:,i_target] + # vars1=Vars[:,i_target] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars1,axis=1)),cmap='viridis') + + plt.figure(figsize=(10,4)) + # plt.subplot(1,3,1) + # plt.plot(vars0,c='r') + # plt.plot(vars1,c='b',alpha=0.6) + # plt.xlabel("time frame") + # plt.ylabel("uncertainty") + # plt.ylim([0,1]) + i_show=123 + plt.subplot(1,3,2) + plt.plot(Contribs[:,i_show],c='r') + # plt.plot(Contribs[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution") + plt.xlabel("time frame") + plt.subplot(1,3,3) + plt.plot(Contribs_sort[:,i_show],c='r') + # plt.plot(Contribs_sort[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution sort") + plt.xlabel("time frame") + plt.show() + + + # [2] Graph initialization for key Gaussians + # Initializations + # 0. naive ori distance + Oris_rel=calculate_edge_relative_orientation(Oris_xyzw) + Ori_distances_naive=get_navie_ori_distances(Oris_rel) # (nk, nk) + + # 1. vertices + # 1.1 positions + Transls=torch.tensor(Transls,dtype=torch.float32) + Transls_optimizable=Transls.clone().requires_grad_(True) + # Transls_optimizable=torch.tensor(Transls,dtype=torch.float64,requires_grad=True) + # Transls_optimizable=torch.nn.Parameter(Transls_optimizable) + # condition1a=Contribs>0.4*np.max(Contribs,axis=0) # 0.707 + # # condition1b=Contribs2>contrib_threshold + # # condition1=np.logical_or(condition1a,condition1b) + # condition1=condition1a + # Mask=torch.tensor(condition1,dtype=torch.bool) + Mask=torch.tensor(Contribs>contrib_threshold,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable=Oris_xyzw.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edge distance + Dists = np.array([distance_matrix(Transls[t], Transls[t]) for t in range(nt)]) + Dists_optimizable=np.mean(Dists,axis=0) + Dists_optimizable=torch.tensor(Dists_optimizable,dtype=torch.float32,requires_grad=True) + + Ori_distances_optimizable=Ori_distances_naive.clone() + Ori_distances_optimizable=torch.tensor(Ori_distances_optimizable,dtype=torch.float32,requires_grad=True) + + # 3. edge weight + weight_distance_optimizable=1*torch.ones(Dists_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_distance_optimizable=weight_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # use the different weight + weight_ori_distance_optimizable=1*torch.ones(Ori_distances_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_ori_distance_optimizable=weight_ori_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # 4. get local edges + n_neighbors=120 + if version_key_edges=="max_contrib": + edges_graph=get_knn_edges(n_neighbors,Transls,Contribs,Transls,Contribs,contrib_threshold,flag_remove_self=True) # edge version: max contrib version + elif version_key_edges=="minmax_distance_contrib": + # edges_graph,_=get_knn_edges_others(Transls,Mask,Transls,Mask,k=n_neighbors) # edge version: minmax distance + contrib mask version + edges_graph,_=get_knn_edges_key(Transls,Mask,k=n_neighbors) + elif version_key_edges=="max_native": + edges_graph,_=get_knn_edges_naive(n_neighbors,Transls,Transls,flag_remove_self=True) # edge version: max contrib version + # breakpoint() + else: + breakpoint() + raise ValueError("version_key_edges should be max_contrib or minmax_distance_contrib or max_native") + # edges_graph.shape # [12395, 2] + weight_distance_optimizable=torch.zeros(Dists_optimizable.shape,dtype=torch.float32) + weight_distance_optimizable[edges_graph[:,0],edges_graph[:,1]]=1 #1e-5 + weight_distance_optimizable.requires_grad=True + + # [3] optimization for key Guassians + print("===========================================") + print("Start Optimization for key Guassians") + print("-------------------------------------------") + + torch.cuda.empty_cache() + Transls_optimized,Dists_optimized,weight_distance_optimized,Oris_xyzw_optimized, Ori_distances_optimized=\ + optimize_overall_with_orientation_edges(\ + edges_graph,\ + Transls, Transls_optimizable.clone(), Mask, Dists_optimizable.clone(), weight_distance_optimizable.clone(),\ + Oris_xyzw, Oris_xyzw_optimizable.clone(), Ori_distances_optimizable.clone(), weight_ori_distance_optimizable, \ + None, None, None,None,\ + W2Cs,Depths_info_perG_key) + + print("-------------------------------------------") + print("Finish Optimization for key Guassians") + print("===========================================") + + if flag_visulize_sparisity: + # breakpoint() + weight_distance_optimized[2,:10],weight_distance_optimized[:10,2] # check symetry + # look at normed weight + torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None] + data = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=50) # check dist + # np.sort(data[4])[-100:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + # (data>1e-6).sum() + np.histogram((data>1e-6).sum(axis=-1)) + # ((data>1e-6).sum(axis=-1)==3).sum() + + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + # Example 2D array + print((data>0.02).sum()) + + # Create a meshgrid for the X and Y coordinates + x = np.arange(data.shape[1]) + y = np.arange(data.shape[0]) + x, y = np.meshgrid(x, y) + + # Create a figure and a 3D axis + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the surface + ax.plot_surface(x, y, np.flip(data,axis=1), cmap='viridis') + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_yticks([0,100,200,300,400]) + ax.set_yticklabels(nk-np.arange(0,500,100)) + ax.set_zlabel('Weight Values') + # ax.set_zlim(0,0.5) + ax.set_title('Edge Weights') + + plt.show() + + + if flag_visulize_key_point: + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=Transls.shape[0]-1 # 0 35 80 100 150 + # t=33 # test backpack + # t=0 # 306 block + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=2) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=2) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + # save + # get optimized graph from normed weights + norm_weight_distance_optimized = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() + edges_graph_optimized=[] + for i in range(nk): + for j in range(nk): + if norm_weight_distance_optimized[i,j]>1e-6: + edges_graph_optimized.append([j,i]) + edges_graph_optimized=torch.tensor(edges_graph_optimized,dtype=torch.long) + + Ori_opt=Oris_xyzw_optimized + Nodes_opt=torch.concatenate([Transls_optimized.transpose(1,0),Ori_opt.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep=[[] for i in range(nk)] + # for e in edges_opt: + for e in edges_graph_optimized: + # Edges_keep[e[0]].append([e[1],e[0]]) + Edges_keep[e[1]].append([e[0],e[1]]) + + Edges_keep=[torch.tensor(Edges_keep[i],device=device) for i in range(nk)] + num_Edges_keep=[Edges_keep[i].shape for i in range(nk)] + # num_Edges_keep + + # check + torch.norm(Oris_xyzw_optimized[0],dim=-1)[:10] + + # for save + optimized_key_output={"Transls_optimized":Transls_optimized,\ + "Dists_optimized":Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized} + + # Save variables to a file + # torch.save({ + # 'dicts': dicts,\ + # 'ids': ids,\ + # 'Contribs': Contribs,\ + # 'Transls': Transls,\ + # 'Oris_xyzw': Oris_xyzw,\ + # 'Transls_optimizable': Transls_optimizable,\ + # 'Dists_optimizable': Dists_optimizable,\ + # 'weight_distance_optimizable':weight_distance_optimizable,\ + # 'Ori_distances_optimizable':Ori_distances_optimizable,\ + # 'Transls_optimized': Transls_optimized,\ + # 'Dists_optimized': Dists_optimized,\ + # "weight_distance_optimized":weight_distance_optimized,\ + # "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + # "Ori_distances_optimized":Ori_distances_optimized,\ + # "edges_graph":edges_graph,\ + # "edges_graph_optimized":edges_graph_optimized,\ + # "Nodes_opt":Nodes_opt,\ + # "Edges_keep":Edges_keep,\ + # "optimized_key_output":optimized_key_output,\ + # # "Transls_optimized1":Transls_optimized1,\ + # # "Dists_optimized1":Dists_optimized1,\ + # # "weight_distance_optimized1":weight_distance_optimized1,\ + # "device":device,\ + # "work_dir":work_dir}\ + + # , 'variables_8.pth') + + flag_read=False + if flag_read: + # Read variables from the file + checkpoint = torch.load('variables_8.pth') + dicts = checkpoint['dicts'] + ids = checkpoint['ids'] + Contribs = checkpoint['Contribs'] + Transls = checkpoint['Transls'] + Oris_xyzw = checkpoint['Oris_xyzw'] + Transls_optimized = checkpoint['Transls_optimized'] + Dists_optimized = checkpoint['Dists_optimized'] + weight_distance_optimized = checkpoint['weight_distance_optimized'] + Oris_xyzw_optimized = checkpoint['Oris_xyzw_optimized'] + Ori_distances_optimized = checkpoint['Ori_distances_optimized'] + edges_graph = checkpoint['edges_graph'] + edges_graph_optimized=checkpoint['edges_graph_optimized'] + Nodes_opt = checkpoint['Nodes_opt'] + Edges_keep = checkpoint['Edges_keep'] + optimized_key_output = checkpoint['optimized_key_output'] + # Transls_optimized1 = checkpoint['Transls_optimized1'] + # Dists_optimized1 = checkpoint['Dists_optimized1'] + # weight_distance_optimized1 = checkpoint['weight_distance_optimized1'] + device=checkpoint['device'] + work_dir=checkpoint['work_dir'] + nk=len(ids) + + + # [4] Graph initialization for non-key Gaussians + # 1. prepare data for non-key Gaussians + Ng=dicts[0]["means"].shape[0] # num of all Gaussians + nt=len(dicts) + # Ng=1000 + ids2=[i for i in range(Ng) if i not in ids] # ids of non-target nodes + # Contribs2_temp=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2_temp=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2_temp=Contribs_all[:,ids2] + Contribs2_temp_max=Contribs2_temp.max(axis=0) + if nt>350: + n_nonkey_max=36000 # 36000 # 150000 72000 too large memory for block teddy wheel + elif nt>40: + n_nonkey_max=72000 + else: + n_nonkey_max=150000 + if Contribs2_temp_max.shape[0]>n_nonkey_max: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[-n_nonkey_max] # the nth (12000th/24000th) largest contrib value + else: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[0] + if nt>40: # iphone davis + contribs2_nth_largest=max(contribs2_nth_largest,0.01) # 0.1 + else: # nvidia + contribs2_nth_largest=max(contribs2_nth_largest,1e-4) + threshold=contribs2_nth_largest # rewrite threshold + ids2=np.array(ids2)[Contribs2_temp.max(axis=0)>threshold].tolist() # remove nodes with small contribs #15 + + + Transls2=np.array([dicts[t]["means"][ids2].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz2=np.array([dicts[t]["quats"][ids2].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw2=Quats_wxyz2[...,[1,2,3,0]] # xyzw + # Contribs2=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2=Contribs_all[:,ids2] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort2=np.array([get_order(Contribs2[:,i]) for i in range(Contribs2.shape[1])]).transpose(1,0) + # Vars2=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # W2Cs2=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + # SE3s2=np.concatenate([Transls2,Quats_xyzw2],axis=-1) # xyz xyzw (required by pypose) + # SE3s2=torch.tensor(SE3s2,dtype=torch.float32) #,device='cuda') + Oris_xyzw2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type) + Oris2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type).Log() + Depths_info_perG_nonkey=Depths_info_perG[:,ids2] + nu=len(ids2) # num of non-target nodes + print(f'Ng={Ng},nk=len(ids)={len(ids)},nu=len(ids2)={len(ids2)},threshold={threshold}') + + # # test threshold + # condition1=Contribs2>0.707*np.max(Contribs2,axis=0) + # condition2=Contribs2>contrib_threshold + # np.histogram((np.logical_or(condition1,condition1)).sum(axis=0),bins=30) + + # 2. run optimizaiton for non-key Guassians + print("===========================================") + print("Start Optimization for nonkey Guassians") + print("-------------------------------------------") + + Transls_optimized2,Dists_optimized2,weight_distance_optimized2,\ + Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others, \ + weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict \ + =run_batch(Transls2,Oris_xyzw2,Contribs2,\ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + Contribs,contrib_threshold,\ + W2Cs,Depths_info_perG_nonkey,\ + Mask) + + print("weight_param_key_optimized2 hist:",np.histogram(weight_param_key_optimized2.cpu().detach().numpy(),bins=10)) + + print("Finish Optimization for nonkey Guassians") + print("===========================================") + + + + if flag_visulize_nonkey_point: + # 3. check + # just for Method 1 + data = (torch.abs(weight_distance_optimized2)/torch.norm(weight_distance_optimized2,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=5) # check dist + np.sort(data[4])[-5:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + np.histogram((data>1e-6).sum(axis=-1)) # check sparsity + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=nt-1 # 0 35 80 100 150 + # t = 168 # backpack test + # t = 0 # 306 block # 33 + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + pts_others=Transls_optimized2.cpu().detach().numpy()[t] + # pts_others=Transls2.cpu().detach().numpy()[t] + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + x3 = pts_others[:,0] + y3 = pts_others[:,1] + z3 = pts_others[:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=4) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=4) + + ax.scatter(x3, y3, z3, c='k', label='other Points',s=1) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + Ori_opt2=torch.tensor(Quats_xyzw2,device=device) # tempory + # Ori_opt2=Oris_xyzw_optimized2.tensor() + Nodes_opt2=torch.concatenate([Transls_optimized2.transpose(1,0),Ori_opt2.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep2=[[] for i in range(nu)] + # for e in edges_opt: + for e in edges_others_optimized: + Edges_keep2[e[0]].append([e[1],e[0]+nk]) # [key id, other id + offset (nk)] + # Edges_keep2[e[1]].append([e[0],e[1]]) + + Edges_keep2=[torch.tensor(Edges_keep2[i],device=device) for i in range(nu)] + num_Edges_keep2=[Edges_keep2[i].shape for i in range(nu)] + + # combine key and others + Nodes_opt_combine=torch.cat([Nodes_opt,Nodes_opt2],dim=0) + Edges_keep_combine=Edges_keep+Edges_keep2 + ids_combine=ids+ids2 + + optimized_others_output={"Transls_optimized":Transls_optimized2,\ + "Dists_optimized":Dists_optimized2,\ + "weight_distance_optimized":weight_distance_optimized2,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized2, \ + "Ori_distances_optimized":Ori_distances_optimized2} + + + # [5] Save graph preprocessing data + quats_xyzw=Nodes_opt[:,:,3:] # xyzw + quats_wxyz=quats_xyzw[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt=[Nodes_opt[:,:,:3],quats_wxyz] # [[nk,nt,3], [nk,nt,4]] + + # combine + quats_xyzw_combine=Nodes_opt_combine[:,:,3:] # xyzw + quats_xyzw_combine=quats_xyzw_combine[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt_combine=[Nodes_opt_combine[:,:,:3],quats_xyzw_combine] # [[nk,nt,3], [nk,nt,4]] + + # tracks_SE3_opt[1].shape,Nodes_opt.shape,tracks_SE3_opt_combine[1].shape,Nodes_opt_combine.shape + + nk=len(Edges_keep) + adjacent_matrix=torch.zeros((nk,nk)) + for i in range(nk): + if Edges_keep[i].shape[0]==0: + continue + indices=Edges_keep[i][:,0] + adjacent_matrix[i,indices]=1 + + adjacent_matrix = torch.logical_or(adjacent_matrix.t(), adjacent_matrix).type(torch.int32) + + Edges_keep_bi=[] + for i in range(nk): + indices=torch.where(adjacent_matrix[i,:]==1)[0] + edges_keep_bi=torch.cat((indices.unsqueeze(0),i*torch.ones_like(indices).unsqueeze(0))).t() + Edges_keep_bi.append(edges_keep_bi) + + Edges_keep_bi_combine=Edges_keep_bi+Edges_keep2 + + + # Save 1: + is_target=list(range(len(ids))) + is_target_combine=list(range(len(ids_combine))) + opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + "optimized_key_output":optimized_key_output,\ + "Nodes_opt_combine":Nodes_opt_combine,"tracks_SE3_opt_combine":tracks_SE3_opt_combine,"ids_combine":np.array(ids_combine),\ + "is_target_combine":is_target_combine,"Edges_keep_combine":Edges_keep_combine,"Edges_keep_bi_combine":Edges_keep_bi_combine,\ + "optimized_others_output":optimized_others_output} + # opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + # "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + # "optimized_key_output":optimized_key_output} + import pickle + dict_dir=f"{work_dir}/opt_dicts" + os.makedirs(dict_dir, exist_ok=True) + file_name=f"opt_v8_{len(ids)}_0.pkl" + # file_name=f"opt_427_var0.002_cut0.002_2.pkl" # 854 eff file + # file_name=f"opt_854_var0.002_cut0.002_nearby.pkl" # 854 eff file + # file_name=f"opt_2544_var0.002_cut0.002.pkl" # 2544 eff file + # file_name=f"opt_2038_var0.002_cut0.002_test.pkl" # 2544 eff file + # file_name="opt_dicts_test3.pkl" + + # # save + # file=open(f'{dict_dir}/{file_name}', 'wb') + # pickle.dump(opt_dict, file) + # file.close() + # print("file_path:",f'{dict_dir}/{file_name}') + + # Save 2: + # Save variables to a file + dict_dir_graph=f"{work_dir}/graph_model/" + os.makedirs(dict_dir_graph, exist_ok=True) + # file_name_graph=f"v8_712_parameters0.pth" + file_path_graph=dict_dir_graph+file_name_graph + + torch.save({ + 'dicts': dicts,\ + 'ids': ids,\ + 'Contribs': Contribs,\ + 'Transls': Transls,\ + 'Oris_xyzw': Oris_xyzw,\ + 'Transls_optimizable': Transls_optimizable,\ + 'Dists_optimizable': Dists_optimizable,\ + 'weight_distance_optimizable':weight_distance_optimizable,\ + 'Ori_distances_optimizable':Ori_distances_optimizable,\ + 'Transls_optimized': Transls_optimized,\ + 'Dists_optimized': Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized,\ + "edges_graph": edges_graph,\ + "edges_graph_optimized": edges_graph_optimized,\ + "Nodes_opt":Nodes_opt,\ + "Edges_keep":Edges_keep,\ + "optimized_key_output":optimized_key_output,\ + # "Transls_optimized1":Transls_optimized1,\ + # "Dists_optimized1":Dists_optimized1,\ + # "weight_distance_optimized1":weight_distance_optimized1,\ + "device":device,\ + "work_dir":work_dir,\ + "ids2":ids2,\ + "Contribs2": Contribs2,\ + "Transls_optimized2":Transls_optimized2,\ + "Dists_optimized2":Dists_optimized2,\ + "weight_distance_optimized2":weight_distance_optimized2,\ + "Oris_xyzw_optimized2":Oris_xyzw_optimized2,\ + "Ori_distances_optimized2":Ori_distances_optimized2,\ + "edges_others_optimized":edges_others_optimized,\ + "edges_others":edges_others,\ + "blend_opt_dict":blend_opt_dict,\ + "contrib_threshold":contrib_threshold,\ + }\ + , file_path_graph) + + print("file_path_graph:",file_path_graph) + + +def main2(save_dir,voxel_size=0.4,n_min_effective_frames=5,ratio_key=-1): + flag_remove_outlier=False + flag_show_sample=False + flag_show_threshold=False + flag_ablation_select_pts_wo_uncertainty=False + flag_ablation_random_select=False + # ratio_key=0.08 # ablation study of threshold and selected number of pts at each step. + + + # folder_dict="/data/dataset/used_ds/som/sriracha-tree/work-dir/out_dicts2" # haru-sit, backpack, sriracha-tree + # folder_dict="/data/dataset/custom_ds_graph_model/rhino/out_dicts3" # rhino + folder_dict=f"{save_dir}/out_dicts3" + + nt=count_out_dict_files(folder_dict) + dicts=[] + for t in range(nt): + with open(f"{folder_dict}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy() for t in range(nt)]) #### (old) + Contribs=np.array([dicts[t]['contribs_strict'].cpu().numpy() for t in range(nt)]) #### + + # Safety check for empty Contribs + if Contribs.size == 0: + breakpoint() + print("Warning: Contribs array is empty, using contribs_in_mask as fallback") + + if ratio_key>0: + max_num_key_nodes=int(Contribs.shape[1]*ratio_key) + num_stage_wise_select_key_nodes=int(max_num_key_nodes/1.5) + max_num_key_nodes = min(max_num_key_nodes, 1500) # avoid used memory is too large. + max_num_key_nodes = max(max_num_key_nodes, 5) # avoid no of key nodes too few. + else: + max_num_key_nodes=1500 # preivous setting + num_stage_wise_select_key_nodes=1000 # previous setting + + print("******ratio_key: ",ratio_key, "max_num_key_nodes: ",max_num_key_nodes," num_stage_wise_select_key_nodes: ",num_stage_wise_select_key_nodes) + + + c99=np.quantile(Contribs, 0.99) + c80=np.quantile(Contribs, 0.80) + contrib_ref=Contribs[(Contribsc80)].mean() + if contrib_ref<0.5: + prod_num=0.5/contrib_ref + Contribs=Contribs*prod_num + Contribs = np.clip(Contribs, a_min=None, a_max=10) + # breakpoint() + # dicts[0].keys(),nt,dicts[0]['means'].shape + + if flag_remove_outlier: + def get_outliers_by_density(means): + X=means[np.arange(0,means.shape[0],1),:] + + # Apply HDBSCAN clustering + # clusterer = hdbscan.HDBSCAN(min_cluster_size=10, min_samples=5, allow_single_cluster=True) + # labels = clusterer.fit_predict(X) + clusterer = DBSCAN(eps=0.02, min_samples=10) ############### hyperparameter!!! + labels=clusterer.fit(X).labels_ + outlier_map=labels==-1 + + # Unique clusters + # n_clusters = len(set(labels)) - (1 if -1 in labels else 0) # Ignore noise label (-1) + # print(f"Detected Clusters: {n_clusters}") + # assert n_clusters==1 + + return outlier_map + + outlier_map=get_outliers_by_density(dicts[t]['means'].cpu().numpy()) + outlier_map.sum() + + Contribs_up=[] + for t in range(nt): + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + if flag_remove_outlier: + outlier_map=get_outliers_by_density(means) + contribs[outlier_map]=1e-10 ## + Contribs_up.append(contribs) + Contribs_up=np.array(Contribs_up) + else: + Contribs_up=None + + print("num of gaussians: ",dicts[0]['means'].shape[0]) + # 1st Gaussian selection round + # select according to contribs + selected_ids_ts=[] + for t in range(nt): + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + else: + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) # ascending order w.r.t. ids + n=contribs.shape[0] + # candidate_ids_map=sorted_indices>=n-1000 + candidates_in_raw_ids=sorted_indices[-num_stage_wise_select_key_nodes:] #1000 + # candidates_in_raw_ids=[i for i in range(n) if candidate_ids_map[i]] + + candidate_means=means[candidates_in_raw_ids] + # candidate_contribs=contribs[candidate_ids_map] + + # voxel_size = 0.4 # 0.5 # 0.4 + voxel_grid = np.floor(candidate_means / voxel_size) + _, unique_indices = np.unique(voxel_grid, axis=0, return_index=True) + selected_candidates_in_raw_ids=[candidates_in_raw_ids[i] for i in unique_indices] + if flag_ablation_random_select: + n_random_select=len(selected_candidates_in_raw_ids) + selected_candidates_in_raw_ids=np.random.choice(selected_candidates_in_raw_ids, n_random_select, replace=False).tolist() + selected_ids_ts.append(selected_candidates_in_raw_ids) + # print(f"t={t}, selected_candidates_in_raw_ids.shape={len(selected_candidates_in_raw_ids)}") + + counts=np.zeros(n) + for t in range(nt): + selected_candidates_in_raw_ids=selected_ids_ts[t] + counts[selected_candidates_in_raw_ids]+=1 + + selected_ids_effective_1frame=[i for i in range(n) if counts[i]>0] + print("#pts left after 1st selection round: ", len(selected_ids_effective_1frame)) + # counts[selected_ids_effective_1frame] + + # avoid too many point in the 1st selection round + # select pts + def select_pts(n,selected_ids_effective_1frame): + arr1=np.ones(n) + arr2=np.zeros(len(selected_ids_effective_1frame)-n) + arr=np.concatenate((arr1,arr2),axis=0) + np.random.shuffle(arr) + selected_ids_effective_1frame=np.array(selected_ids_effective_1frame)[arr.astype(bool)].tolist() + return selected_ids_effective_1frame + # if len(selected_ids_effective_1frame)>2200: + # selected_ids_effective_1frame=select_pts(2200,selected_ids_effective_1frame) + # print("#pts left after 1st selection round (adjusted): ", len(selected_ids_effective_1frame)) + + # 2nd Gaussian selection round + ## 2nd selection round: whose contribs are effective at least nt/6 frames + # n_min_effective_frames=30 # 30 + # n_min_effective_frames=int(nt/6) + # n_min_effective_frames=min(5,int(nt/6)) # use 30 to avoid key pts not find in the long video. # TODO # 5 for haru-sit + print("n_min_effective_frames: ",n_min_effective_frames) + counts_effective_frames=np.zeros(len(selected_ids_effective_1frame)) + thresholds=[] + for t in range(nt): + if t%100==0: + print(t) + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + else: + # means=dicts[t]['means'].cpu().numpy() + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) + # candidates_in_raw_ids=sorted_indices[-2000:] + threshold=contribs[sorted_indices[-num_stage_wise_select_key_nodes]] # the 2000th largest contrib # 1000 + threshold=max(threshold,0.5) # 0.5 + counts_effective_frames+=(contribs[selected_ids_effective_1frame]>=threshold).astype(int) + thresholds.append(threshold) + + threshold_min=np.min(np.array(thresholds)) + threshold_mean=np.mean(np.array(thresholds)) + print("min contrib threshold: ",threshold_min) + print("mean contrib threshold: ",threshold_mean) + print(np.histogram(counts_effective_frames)) + if flag_show_threshold: + import matplotlib.pyplot as plt + plt.figure() + plt.plot(thresholds) + plt.show() + selected_ids_eff=[selected_ids_effective_1frame[i] for i in range(len(selected_ids_effective_1frame)) if counts_effective_frames[i]>=n_min_effective_frames] + print("#pts left after 2nd selection round: ", len(selected_ids_eff)) + # counts_effective_frames[counts_effective_frames>30].shape + + # avoid too many point in the 2nd selection round + if len(selected_ids_eff)>max_num_key_nodes: # 1000 2500 # 1500 + selected_ids_eff=select_pts(max_num_key_nodes,selected_ids_eff) # 1000 2500 1500 + print("#pts left after 2nd selection round (adjusted): ", len(selected_ids_eff)) + + if flag_ablation_select_pts_wo_uncertainty: + selected_ids_eff=np.arange(Contribs.shape[1]) + selected_ids_eff=select_pts(1000,selected_ids_eff) # 1000 # 2500 1500 + print(f"flag_ablation_select_pts_wo_uncertainty={flag_ablation_select_pts_wo_uncertainty}") + print("random select key pts:", len(selected_ids_eff)) + + if flag_show_sample: + import matplotlib.pyplot as plt + # t=nt-1 + t=0 + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + x = means[selected_ids_eff][:,0] + y = means[selected_ids_eff][:,1] + z = means[selected_ids_eff][:,2] + # sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=np.log(contribs[selected_ids_eff]),cmap='viridis') + sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=contribs[selected_ids_eff],cmap='viridis') + + cbar = plt.colorbar(sc) + cbar.set_label('Contribs') + ax.set_xlabel('X Label') + ax.set_ylabel('Y Label') + ax.set_zlabel('Z Label') + ax.axis('equal') + plt.show() + + # save + file_path=f"{folder_dict}/effective_dict_{len(selected_ids_eff)}.pkl" + effective_dict={"ids_eff2":selected_ids_eff,"ids_eff1":selected_ids_effective_1frame,"Contribs_up":Contribs_up, + "threshold_min":threshold_min,"threshold_mean":threshold_mean,"Contribs":Contribs} + file=open(file_path, 'wb') + pickle.dump(effective_dict, file) + file.close() + print(file_path) + file_effective=f"effective_dict_{len(selected_ids_eff)}.pkl" + return file_effective + +def compute_knn_indices(points, k, chunk_size=1024): + """ + Memory-efficient KNN search: returns indices of k nearest neighbors for each point. + points: [N, 3] + returns: [N, k] LongTensor + """ + N = points.shape[0] + knn_indices = [] + + for i in range(0, N, chunk_size): + end = min(i + chunk_size, N) + chunk = points[i:end] # [chunk, 3] + dists = torch.cdist(chunk, points) # [chunk, N] + topk = dists.topk(k + 1, largest=False).indices[:, 1:] # exclude self + knn_indices.append(topk) + + return torch.cat(knn_indices, dim=0) # [N, k] + +def gaussian_kde_density_knn(points, inv_covariances, k=1000, eps=1e-6): + """ + Vectorized version using k nearest neighbors for KDE density. + points: [N, 3] + covariances: [N, 3, 3] + returns: [N] density using k-NN Mahalanobis distances + """ + N, D = points.shape + device = points.device + + # Invert covariances: [N, 3, 3] + # cov_inv = torch.inverse(covariances + eps * torch.eye(D, device=device)) + cov_inv = inv_covariances + + # # Get k-NN indices using Euclidean distance + # dists = torch.cdist(points, points) # [N, N] + # knn_indices = dists.topk(k + 1, largest=False).indices[:, 1:] # [N, k] (exclude self) + + # Memory-efficient KNN + chunk_size= 10000 + knn_indices = compute_knn_indices(points, k, chunk_size=chunk_size) # [N, k] + + # Gather neighbor points: [N, k, 3] + knn_points = points[knn_indices] # batched gather + + # Center point per row: [N, 1, 3] -> broadcast to [N, k, 3] + diffs = knn_points - points[:, None, :] # [N, k, 3] + + # Apply Mahalanobis: dᵢⱼ = (xᵢ - xⱼ)^T Σᵢ⁻¹ (xᵢ - xⱼ) + # cov_inv: [N, 3, 3], diffs: [N, k, 3] + # compute: diffᵢⱼ^T @ cov_invᵢ @ diffᵢⱼ -> result [N, k] + + diffs_unsq = diffs.unsqueeze(-1) # [N, k, 3, 1] + cov_inv_exp = cov_inv[:, None, :, :] # [N, 1, 3, 3] + m = torch.matmul(cov_inv_exp, diffs_unsq) # [N, k, 3, 1] + mahalanobis_sq = torch.matmul(diffs_unsq.transpose(-1, -2), m).squeeze(-1).squeeze(-1) # [N, k] + + # Gaussian kernel weights + weights = torch.exp(-0.5 * mahalanobis_sq) # [N, k] + + density = weights.sum(dim=1) # [N] + return density + + +def plot_density_histogram_rgb(density, bins=20, height=256): + """ + density: [N] tensor + height: desired output image height in pixels + Returns: [height, width, 3] RGB uint8 numpy image of the histogram with log-scale x-axis and labeled axes + """ + density_np = density.cpu().numpy() + density_np = density_np[density_np > 0] # filter out non-positive for log scale + + dpi = 100 # Dots per inch + fig_height_inches = height / dpi + aspect_ratio = 1.5 # width / height + fig_width_inches = fig_height_inches * aspect_ratio + + # Create figure with desired size + fig, ax = plt.subplots(figsize=(fig_width_inches, fig_height_inches), dpi=dpi) + ax.hist(density_np, bins=bins, color='skyblue', edgecolor='black') + ax.set_xscale("log") + ax.set_title("Density Histogram (Log X-axis)") + ax.set_xlabel("Density (log scale)") + ax.set_ylabel("Count") + + fig.tight_layout() + fig.canvas.draw() + + width, height = fig.canvas.get_width_height() + image = np.frombuffer(fig.canvas.tostring_rgb(), dtype=np.uint8).reshape(height, width, 3) + + plt.close(fig) + return image.astype(np.uint8) + +def load_gt_masks(ws, gt_mask_dirname="mask"): + import imageio.v2 as imageio + """Load ground truth masks from directory""" + gt_mask_dir = osp.join(ws, gt_mask_dirname) + if not osp.exists(gt_mask_dir): + logging.warning(f"GT mask directory {gt_mask_dir} not found") + breakpoint() + return None + + gt_mask_fns = [ + f for f in os.listdir(gt_mask_dir) + if f.endswith(".png") or f.endswith(".jpg") + ] + gt_mask_fns.sort() + + gt_masks = [] + for gt_mask_fn in gt_mask_fns: + gt_mask_path = osp.join(gt_mask_dir, gt_mask_fn) + gt_mask = imageio.imread(gt_mask_path) + if gt_mask.ndim == 3: + gt_mask = gt_mask[..., 0] # Take first channel if RGB + gt_mask = (gt_mask > 128).astype(np.float32) # Convert to binary + gt_masks.append(gt_mask) + + gt_masks = torch.Tensor(np.stack(gt_masks)).float() # T,H,W + # self.register_gradfree_buffer("gt_masks", gt_masks) + # self.has_gt_masks = True + logging.info(f"Loaded GT masks from {gt_mask_dir}: {gt_masks.shape}, foreground ratio: {gt_masks.mean():.3f}") + return gt_masks, gt_mask_fns + +# @torch.inference_mode() +def main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model=None): + import torch + import argparse + import os + import torch + import time + from leo_renderer import Renderer + from lib_render.render_helper import render, render_gsplat + import numpy as np + from tqdm import tqdm + import imageio.v3 as iio + + from datetime import datetime + from loguru import logger as guru + import torch.nn.functional as F + import cv2 + import matplotlib.pyplot as plt + import lpips + + from lib_ugraph.ugraph_utils import compute_ssim_map, compute_lpips_map + + # parser = argparse.ArgumentParser() + # parser.add_argument("--cfg_fn", type=str) + # parser.add_argument("--port", type=int, default=8899) + # parser.add_argument("--work_dir", type=str, required=True) + # args = parser.parse_args() + + + + if dataset_name=='iphone': + cfg_fn='/data/repo/MoSca/profile/iphone/iphone_fit.yaml' + elif dataset_name=='davis' or dataset_name=='davis_mask': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='nvidia': + cfg_fn='/data/repo/MoSca/profile/nvidia/nvidia_fit.yaml' + elif dataset_name=='diffusion4d': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='objaverse': + cfg_fn='/data/repo/MoSca/profile/objaverse/objaverse_fit.yaml' + else: + cfg_fn='' + breakpoint() + port=9000 + + device = torch.device("cuda") # if torch.cuda.is_available() else "cpu") + renderer = Renderer(cfg_fn, device, work_dir, port=port, bg_flag=True, fg_flag=True,load_s2d=True, ws=ws, use_ugraph=False, relative_dir_saved_mosca_model=relative_dir_saved_mosca_model) + + nt=renderer.cams.T + print("nt:",nt) + + W, H = renderer.cams.default_W, renderer.cams.default_H + assert W == renderer.s2d.W and H == renderer.s2d.H + K = renderer.cams.K(H=H,W=W) + # focal = 0.5 * H / np.tan(0.5 * camera_state.fov).item() + # K = torch.tensor( + # [[focal, 0.0, W / 2.0], [0.0, focal, H / 2.0], [0.0, 0.0, 1.0]], + # device=device, + # ) + # w2c = torch.linalg.inv( + # torch.from_numpy(camera_state.c2w.astype(np.float32)).to(self.device) + # ) + ts=torch.arange(nt,device=device) + w2cs=torch.stack([renderer.cams.T_cw(t) for t in ts],dim=0) + # with torch.inference_mode(): + # for w2c,t in tqdm(zip(w2cs,ts)): + # # w2c = renderer.cams.T_cw(t) + + # gs5 = [] + # if renderer.bg_flag: + # gs5.append(renderer.s_model()) + # if renderer.fg_flag: + # gs5.append(renderer.d_model(t)) + # render_dict, _ = render_gsplat( + # gs5, + # H, + # W, + # K=K, + # T_cw=w2c, + # bg_color=[1.0, 1.0, 1.0], + # ) + + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # # render_dict0 = render( + # # gs5, + # # H, + # # W, + # # K=K, + # # T_cw=w2c, + # # bg_color=[1.0, 1.0, 1.0], + # # ) + + # # img0 = torch.clamp(render_dict0["rgb"].permute(1, 2, 0), 0.0, 1.0) + # # img0 = (img0.cpu().numpy() * 255.0).astype(np.uint8) + # # combined = np.hstack([img0,img]) + # # iio.imwrite(f"./test/img_{t.item()}.png", combined) # check if they look the same. + # # compare_images(img, img0) + + if save_dir=='': + save_dir = work_dir + + video_dir = f"{save_dir}/videos/{datetime.now().strftime('%Y-%m-%d-%H%M%S')}" + print("*"*20) + print("*"*20) + print(f"Saving video to {video_dir}") + print("*"*20) + print("*"*20) + os.makedirs(video_dir, exist_ok=True) + + torch.backends.cuda.preferred_linalg_library("magma") + + # if dataset_name=="davis": + # prior_depths=torch.stack(train_dataset.depths).to(device) + # elif dataset_name=="iphone": + # prior_depths=train_dataset.depths.to(device) # [nt,H,W] + # else: + # # breakpoint() + # print("No prior depth!!!") + + if dataset_name=="davis" or dataset_name=="iphone" or dataset_name=="diffusion4d" or dataset_name=="objaverse" or dataset_name=="davis_mask": + thr_density_strict=0.1 #5 + thr_density_loose=0.1 #1 + thr_rgb_diff=0.3 + thr_ssim=0.01 + elif dataset_name=="nvidia": + thr_density_strict=0.0 + thr_density_loose=0.0 + thr_rgb_diff=0.6#0.6 + thr_ssim=0.001 # 0.001 + else: + breakpoint() + + try: + # gt_mask_dir = osp.join(ws, "mask") + # renderer.s2d.load_gt_masks(gt_mask_dir) + gt_masks=renderer.s2d.gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + except: + breakpoint() + # gt_masks=None + gt_masks, gt_mask_fns=load_gt_masks(ws, "mask") + gt_masks=gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + # breakpoint() + prior_depths=renderer.s2d.dep.to(device) # [nt,H,W] + # breakpoint() + prior_depths_grad, _, _=compute_depth_gradient(prior_depths) # [nt,H,W] + # depths_info = 0.05*torch.exp(-30*prior_depths_grad) # [nt,H,W] # TODO: hyperparameter + # depths_info = depths_info.clamp_min(1e-4) # set minimum value to 1e-4 + depths_info = 1e-3 * torch.ones_like(prior_depths_grad,device=device) # 1e-3 1e-5 + # depths_info = 1e-3 * torch.ones((nt,H,W),device=device) # 1e-3 1e-5 + # depths_info[prior_depths_grad <= 0.03] = 0.1 # [nt,H,W] # TODO: hyperparameter # 0.05 # TODO!!! + print("np.histogram:",np.histogram(prior_depths_grad[gt_masks].flatten().cpu().numpy(), bins=10)) + # np.histogram(depths_info.flatten().cpu().numpy(), bins=10) + depth_min = prior_depths.min().item() + def safe_quantile(tensor: torch.Tensor, q: float, max_samples: int = 100000): + """Compute quantile safely by subsampling if needed.""" + if tensor.numel() > max_samples: + indices = torch.randperm(tensor.numel(), device=tensor.device)[:max_samples] + tensor = tensor.view(-1)[indices] + return tensor.quantile(q) + depth_max = safe_quantile(prior_depths, 0.99).item() + prior_depths_grad_threshold = safe_quantile(prior_depths_grad[gt_masks], 0.90).item() + print(f"set prior_depths_grad_threshold to {prior_depths_grad_threshold}.") + # lpips_model = lpips.LPIPS(net="alex", spatial=True).to(device) + # breakpoint() + print("fg Gaussian num:", renderer.d_model.N) + print("bg Gaussian num:", renderer.s_model.N) + print("generate out_dicts_t.pkl:") + for i, (w2c, t) in enumerate(zip(tqdm(w2cs), ts)): + assert i == t.item() + # if i<98: + # continue # test + with torch.inference_mode(): + # out_dict=renderer.model.render(int(t.item()), w2c[None], K[None], img_wh) # my code + gs5 = [] + if renderer.bg_flag: + gs5.append(renderer.s_model()) + if renderer.fg_flag: + d_list_t=renderer.d_model(t) + gs5.append(d_list_t) + n_fg = renderer.d_model(t)[0].shape[0] + render_dict, out_dict = render_gsplat( + gs5, + H, + W, + K=K, + T_cw=w2c, + bg_color=[1.0, 1.0, 1.0], + n_fg=n_fg, + ) + n_all=len(gs5) + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # density filter begin + mu, frame, scale, o, sph=d_list_t + # S = torch.zeros_like(frame) + # S[:, 0, 0] = scale[:, 0] + # S[:, 1, 1] = scale[:, 1] + # S[:, 2, 2] = scale[:, 2] + # actual_covariance = frame @ (S**2) @ frame.permute(0, 2, 1) + inv_actual_covariance = frame @ torch.diag_embed(1.0 / (scale ** 2)) @ frame.transpose(1, 2) + + # test = torch.distributions.MultivariateNormal(mu, actual_covariance).log_prob(mu) + density=gaussian_kde_density_knn(mu,inv_actual_covariance, k=100) # TODO: check scale factor + # breakpoint() + large_density_Gaussian_strict=density>thr_density_strict # TODO: threshold # 5 + large_density_Gaussian_loose=density>thr_density_loose # TODO: threshold # 1 + # density filter end + + out_dict['w2c']=w2c.detach() + out_dict['K']=K + # gt_mask = renderer.s2d.dep_mask[i].to(device) # (H, W) # test # not the true mask! + gt_mask = gt_masks[i] + # breakpoint() + # out_dict['gt_mask']=gt_mask + projected = project_points(out_dict['means'], K, w2c) # [N,2] + in_mask_Gaussians=valid_and_visible(projected, gt_mask) # [N] bool + out_dict['in_mask_Gaussians']=in_mask_Gaussians + contribs=out_dict['contribs'].reshape(-1,3)[:,0].clone() + + img = out_dict["img"][0] # [H,W,3] + gt_img = renderer.s2d.rgb[t].to(device) # [H,W,3] + diff_img = F.l1_loss(img, gt_img, reduction="none").mean(dim=-1) # [H,W] + mask_diff_img = diff_img < thr_rgb_diff # [H,W] # TODO: threshold #0.1 + small_img_diff_Gaussians=valid_and_visible(projected, mask_diff_img) # [N] bool + + # ssim map begin + ssim_map, ssim_loss=compute_ssim_map(img.cpu().numpy(),gt_img.cpu().numpy()) # [H,W,3] + mask_ssim_img = torch.tensor(np.array(ssim_map > thr_ssim),device=device).all(dim=-1) # [H,W] # TODO + large_ssim_Gaussians=valid_and_visible(projected, mask_ssim_img) # [N] bool + + # lpips_map, lpips_loss=compute_lpips_map(lpips_model, img.cpu().numpy(),gt_img.cpu().numpy(),device=device) # [H,W,1] + # mask_lpips_img = torch.tensor((lpips_map < 0.5)[:,:,0],device=device) # [H,W] # TODO + # small_lpips_Gaussians=valid_and_visible(projected, mask_lpips_img) # [N] bool + # breakpoint() + # ssim map end + # depth gradient mask + mask_depth_grad_img = prior_depths_grad[i] < prior_depths_grad_threshold # [H,W] + small_depth_grad_Gaussians=valid_and_visible(projected, mask_depth_grad_img) # [N] bool + + # + valid_gaussians_strict = small_depth_grad_Gaussians & large_ssim_Gaussians & small_img_diff_Gaussians & in_mask_Gaussians # & large_density_Gaussian_strict #& small_lpips_Gaussians # & in_mask_Gaussians # + valid_gaussians_loose = large_ssim_Gaussians & small_img_diff_Gaussians # & large_density_Gaussian_loose #& small_lpips_Gaussians # + # valid_gaussians=torch.ones(n_all, dtype=torch.bool, device=device) # TODO: threshold + # contribs_in_mask[~small_img_diff_Gaussians]=1e-8 # Gaussians on incorrect pixel are set to small contribs + contribs_in_mask=contribs.clone() + contribs_in_mask[~in_mask_Gaussians]=1e-8 # out-mask Gaussians are set to small contribs + contribs_strict=contribs.clone() + contribs_loose=contribs.clone() + contribs_strict[~valid_gaussians_strict]=1e-8 + contribs_loose[~valid_gaussians_loose]=1e-8 + + out_dict['contribs_in_mask']=contribs_in_mask + out_dict['contribs_strict']=contribs_strict + out_dict['contribs_loose']=contribs_loose + + # depth uncertainty + depths_info_perG=associate_pts_value(projected, depths_info[i]) # per Gaussian [N] + out_dict['depths_info_perG']=depths_info_perG + # out_dict['prior_depths_grad']=prior_depths_grad + + if True: + import pickle + dict_dir=f"{save_dir}/out_dicts3" + os.makedirs(dict_dir, exist_ok=True) + file=open(f'{dict_dir}/out_dict_{t}.pkl', 'wb') + pickle.dump(out_dict, file) + file.close() + + if True: + def to_rgb(img: np.ndarray) -> np.ndarray: + # [H, W] or [H, W, 1] -> [H, W, 3] + if img.ndim == 2: + return cv2.cvtColor(img, cv2.COLOR_GRAY2BGR) + elif img.ndim == 3 and img.shape[2] == 1: + return cv2.cvtColor(img[:, :, 0], cv2.COLOR_GRAY2BGR) + return img # Already RGB or unsupported format + + img_np = (img.cpu().numpy() * 255.0).astype(np.uint8) + gt_img_np = (gt_img.cpu().numpy() * 255.0).astype(np.uint8) + ssim_map_np = (np.array(ssim_map) * 255.0).clip(0, 255).astype(np.uint8) + # lpips_map_np = to_rgb((np.array(lpips_map) * 255.0).clip(0, 255).astype(np.uint8)) + + img_with_points = draw_projected_points_masked(img_np, projected, valid_gaussians_strict) # valid_gaussians, in_mask_Gaussians + img_with_points_loose = draw_projected_points_masked(img_np, projected, valid_gaussians_loose) + masked_img = overlay_mask_yellow(img_np, gt_mask.cpu().numpy()) + masked_gt_img = overlay_mask_yellow(gt_img_np, gt_mask.cpu().numpy()) + masked_diff_gt_img = overlay_mask_yellow(gt_img_np, mask_diff_img.cpu().numpy(), color= (255,0,0)) + + diff_img_np = ((diff_img*5).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # *10 to make it clear + diff_img_np = to_rgb(diff_img_np) + # depth_info_np= ((100*depths_info[t]).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # TODO: auto normalize for show + # depth_info_np = to_rgb(depth_info_np) + img_with_points_depth=draw_projected_points_colored(img_np, projected, depths_info_perG) + depth_img=apply_depth_colormap( + prior_depths[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = to_rgb(depth_img_np) + prior_depths_grad_img=apply_depth_colormap( + prior_depths_grad[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + prior_depths_grad_img_np = (prior_depths_grad_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + + # hist_img = plot_density_histogram_rgb(density,height=img_np.shape[0]) + + # combined = stack_images_side_by_side(img_with_points, masked_img) + # lpips_map_np, masked_img, masked_gt_img, hist_img , depth_info_np, depth_img_np, + combined = np.hstack([img_with_points, img_with_points_loose, depth_img_np, prior_depths_grad_img_np, ssim_map_np, masked_img, masked_gt_img, masked_diff_gt_img, diff_img_np, img_with_points_depth]) + iio.imwrite(f"{video_dir}/img_{i}.png", combined) + # breakpoint() + # del lpips_model.net + # lpips_model=None + # Delete large objects manually + del renderer + del render_dict + del inv_actual_covariance + # del lpips_model + del prior_depths + del prior_depths_grad + out_dict = {k: v.detach().cpu() if isinstance(v, torch.Tensor) else v for k, v in out_dict.items()} + del out_dict + del img, gt_img, diff_img, mask_diff_img + del projected, depths_info_perG + del contribs_strict, contribs_loose + del ssim_map, ssim_loss#, lpips_map, lpips_loss + + + # Optionally clear lists + gs5.clear() + + # Force garbage collection + import gc + gc.collect() + + # Clear PyTorch CUDA cache + torch.cuda.empty_cache() + # breakpoint() + +def compare_images(img, img0, tolerance=1): + if img.shape != img0.shape: + print(f"Shape mismatch: {img.shape} vs {img0.shape}") + return False + + diff = np.abs(img.astype(np.int32) - img0.astype(np.int32)) + max_diff = diff.max() + + if max_diff <= tolerance: + print(f"Images are close. Max diff: {max_diff}") + return True + else: + print(f"Images differ. Max diff: {max_diff}") + return False + +# def test(): +# # iphone +# # seq_name="block" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five, spin +# seq_name="backpack_" #"backpack_" +# # work_dir=f"/data/dataset/iphone_mosca_trained/paper-windmill/logs/iphone_fit_native_add3_20250226_115418" +# work_dir=f"/data/dataset/iphone_mosca_trained/backpack_/logs/iphone_fit_native_add3" #_20250223_123930" +# save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" +# dataset_name="iphone" + +# file_name_graph=f"v12_parameters8mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) +# main3(save_dir,file_effective,file_name_graph) + +# def test3(): +# # davis +# seq_name="camel" # rhino, camel, car-turn +# work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" +# save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" +# # root_dir=f"/data/dataset/custom_mocsa_initialized" +# dataset_name="davis" + +# file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 +# main3(save_dir,file_effective,file_name_graph) + + +def test2(): + # iphone + # seq_names=['paper-windmill', 'block', 'teddy', 'apple', 'haru-sit']#, 'sriracha-tree', 'mochi-high-five', ] #,'sriracha-tree','block','teddy'] + # seq_name="mochi-high-five" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five + # 'backpack_', + # seq_names=['backpack_', 'apple', 'block', 'handwavy_', 'mochi-high-five_', 'pillow_', 'spin', 'teddy', 'creeper_', 'haru-sit_', 'paper-windmill', 'space-out', 'sriracha-tree_', 'wheel'] + # seq_names=['apple', 'block', 'spin', 'teddy', 'space-out', 'wheel'] + seq_names=['apple'] + + depth_ratio=0.1 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + # auto find the folder with timestamp + # base_dir = f"/data/dataset/iphone_mosca_trained/{seq_name}/logs" + # pattern = os.path.join(base_dir, "iphone_fit_native_add3*") + # matches = glob.glob(pattern) + # if matches: + # work_dir = matches[0] # or choose with more logic + # print("Found:", work_dir) + # else: + # print("No matching folders found.") + # print('='*20) + # continue + + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/logs/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + dataset_name="iphone" + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + # main1(dataset_name,work_dir,save_dir,ws) + torch.cuda.empty_cache() + file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="max_contrib",threshold_min=0.5) #"max_contrib" minmax_distance_contrib + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + + + +def test4(): + # davis # 'car-turn', 'camel', + # seq_names=['car-turn', 'bear', 'breakdance-flare', 'dance-twirl', 'drift-straight', 'helicopter', 'libby', 'mbike-santa', 'rhino', 'schoolgirls'] # + seq_names=['camel'] + # seq_names=['car-turn', 'bear', 'drift-straight', 'helicopter', 'mbike-santa', 'rhino'] # + depth_ratio=0.001 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + dataset_name="davis" + + file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning + # main1(dataset_name,work_dir,save_dir,ws) + file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="minmax_distance_contrib",threshold_min=0.5) #"minmax_distance_contrib") + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + +def backup_code(save_dir): + from datetime import datetime + backup_dir = osp.join(save_dir, "src_backup") + os.makedirs(backup_dir, exist_ok=True) + for path in [ + "run_preprocessing.sh", + "graph_model_preprocessing5.py", + ]: + os.system(f"cp -r {path} {backup_dir}") + # reduce the backup size + # shutil.rmtree(osp.join(backup_dir, "lib_prior", "seg")) + # for root, dirs, files in os.walk(backup_dir): + # for file in files: + # if file.endswith(".pth") or file.endswith(".ckpt"): + # if osp.isfile(osp.join(root, file)): + # os.remove(osp.join(root, file)) + # else: + # shutil.rmtree(osp.join(root, file)) + # backu the commandline args + with open(osp.join(save_dir, f"commandline_args{datetime.now().strftime('%Y-%m-%d-%H%M%S')}.txt"), "w") as f: + f.write(" ".join(sys.argv)) + +def run_from_outside(): + parser = argparse.ArgumentParser() + parser.add_argument("--seq_name", type=str, required=True) + parser.add_argument("--func_name", type=str, required=True) + parser.add_argument("--dataset_name", type=str, required=True) + parser.add_argument("--depth_ratio", type=float, help="default=0.1 (iphone) or 0.001 (backpack or davis)", required=True) # default=0.1, + parser.add_argument("--threshold_min_set", type=float, help="default=0.5 or 0.1 (paper-windmill)", required=True) # default=0.5, + parser.add_argument("--version_key_edges", type=str, help="version_key_edges: max_contrib or minmax_distance_contrib", required=True) # default="max_contrib" + parser.add_argument("--extra_save_str", type=str, help="extra_save_str", required=False, default="") # _ab_uknn ## ablation study + parser.add_argument("--ratio_key", type=float, help="ratio of key nodes to determine the selected number of nodes in the first stage and threshold in the 2nd stage", required=False, default=-1) + parser.add_argument("--which_logs_folder", type=str, help="which_logs_folder", required=False, default="logs") # default="logs" + parser.add_argument("--log_subfolder", type=str, help="log_subfolder", required=False, default="demo_fit_native_add3") # default="demo_fit_native_add3" + # parser.add_argument("--use_ugraph", action="store_true") + parser.add_argument("--relative_dir_saved_mosca_model", type=str, help="relative_dir_saved_mosca_model. e.g.,", required=False) + args = parser.parse_args() + + seq_name=args.seq_name + dataset_name=args.dataset_name + depth_ratio=args.depth_ratio # 0.001 + threshold_min_set=args.threshold_min_set # 0.5 # paper-windmill 0.1 + version_key_edges=args.version_key_edges + extra_save_str=args.extra_save_str + ratio_key=args.ratio_key + # save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}/' # original + save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}{extra_save_str}/' + logs_folder=args.which_logs_folder + relative_dir_saved_mosca_model=args.relative_dir_saved_mosca_model + log_subfolder=args.log_subfolder + if dataset_name=="iphone": + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/{logs_folder}/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + voxel_size=0.2 + n_min_effective_frames=5 + # version_key_edges="max_contrib" + elif dataset_name=="davis": + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # version_key_edges="minmax_distance_contrib" + elif dataset_name=="nvidia": + work_dir=f"/data/dataset/nvidia_mosca_trained/{seq_name}/{logs_folder}/nvidia_fit_native_add3" + save_dir=f"/data/dataset/nvidia_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/nvidia_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=3 + elif dataset_name=="diffusion4d": + work_dir=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/diffusion4d_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # elif dataset_name=="objaverse": # old + # group_name="000-001" + # # /data/dataset/objaverse_mosca_trained/000-001/3ae0b4e6a75a41d38a2adf8db048a703/000/mono/logs/demo_fit_native_add3 + # work_dir=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/{logs_folder}/demo_fit_native_add3" + # save_dir=f"/data/dataset/objaverse_mosca_graph_model/{group_name}/{seq_name}/000/mono/"+save_sub_folder + # ws=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/" + # voxel_size=0.05 + # n_min_effective_frames=5 + elif dataset_name=="objaverse": + work_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/{logs_folder}/{log_subfolder}" # TODO demo_fit_native_add3_0 + save_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_graph_model/{seq_name}/mono/"+save_sub_folder + ws=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/" + voxel_size=0.1 # TODO + n_min_effective_frames=20 # TODO + elif dataset_name=="davis_mask": + work_dir=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model_masked_4x/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + else: + raise ValueError("dataset_name not supported") + + backup_code(save_dir) + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + + if args.func_name=="main1": + main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model) + elif args.func_name=="main23": + if args.depth_ratio is None or args.threshold_min_set is None or args.version_key_edges is None: + parser.error("--depth_ratio, --threshold_min_set, and --version_key_edges are required when func_name is 'main23'") + file_effective=main2(save_dir,voxel_size=voxel_size,n_min_effective_frames=n_min_effective_frames,ratio_key=ratio_key) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges=version_key_edges,threshold_min_set=threshold_min_set) #"max_contrib" minmax_distance_contrib + + + +if __name__ == "__main__": + # test2() + # test4() + # DON'T FORGET TO CHANGE! + run_from_outside() \ No newline at end of file diff --git a/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh new file mode 100644 index 0000000000000000000000000000000000000000..fade29ad45ed691be51bd485f7dd3274bb62fad1 --- /dev/null +++ b/000-033-01dda79d3bd347bdb3326e0586aba1ff-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh @@ -0,0 +1,159 @@ +#!/bin/bash + +gpu_id=3 +dataset_name=iphone + +# 'backpack_' 'haru-sit_' 'paper-windmill' +seq_names=('paper-windmill' 'block' 'teddy' 'wheel' 'apple' 'spin' 'space-out') +# seq_names=('teddy' 'wheel') +# seq_names=('spin' 'space-out') +# seq_names=('paper-windmill') +# seq_names=('block') +# seq_names=('haru-sit_' 'handwavy_' 'mochi-high-five_' 'sriracha-tree_' 'creeper_' 'pillow_') +# seq_names=('haru-sit_') # 'backpack_' + +depth_ratio=0.1 +# threshold_min_set=0.1 # 0.1 paper-windmill # 0.5 others +threshold_min_set_dict=( + ["paper-windmill"]=0.1 + ["teddy"]=0.5 + ["wheel"]=0.5 + ["apple"]=0.5 + ["spin"]=0.5 + ["space-out"]=0.5 + ["haru-sit_"]=0.5 + ["handwavy_"]=0.5 + ["mochi-high-five_"]=0.5 + ["sriracha-tree_"]=0.5 + ["creeper_"]=0.5 + ["pillow_"]=0.5 +) +version_key_edges=max_contrib # max_contrib minmax_distance_contrib max_native +ratio_key=0.02 # default -1 +extra_save_str=_ab_key_ratio_$ratio_key #2 + +# Loop and print each name +for seq_name in "${seq_names[@]}" +do + echo "${seq_name}" +done + +# Loop over each sequence +for seq_name in "${seq_names[@]}" +do + echo "Starting training for ${seq_name}" + + threshold_min_set=${threshold_min_set_dict[$seq_name]} + + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main1 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str # comment if no extra_save_str + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main23 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str \ + --ratio_key $ratio_key # ablation study + + echo "Finished training for ${seq_name}" + echo "*****************************************" + echo "*****************************************" +done + + + + + + +# gpu_id=2 +# dataset_name=davis +# seq_names=('camel' 'car-turn' 'drift-straight' 'helicopter' 'mbike-santa' 'koala' 'parkour' 'sheep' 'soccerball' 'schoolgirls') # 'bear' 'rhino' +# # seq_names=('camel' 'train' 'car-roundabout' 'car-turn' 'bike-packing' 'drift-straight' 'helicopter' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'mbike-santa' 'parkour' 'pigs' 'rhino' 'schoolgirls' 'sheep' 'soccerball' 'swing' 'bear') +# # seq_names=('train' 'car-roundabout' 'bike-packing' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'parkour' 'pigs' 'schoolgirls' 'sheep' 'soccerball' 'swing') + +# depth_ratio=0.001 # 0.1 +# threshold_min_set=0.1 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done + + + +# gpu_id=1 +# dataset_name=nvidia + +# # seq_names=('Balloon1' 'Balloon2' 'Jumping' 'Playground' 'Skating' 'Truck' 'Umbrella') # +# # seq_names=('Balloon1' 'Jumping' 'Truck' ) # 'Umbrella' +# seq_names=('Balloon2' 'Playground' 'Skating') + +# depth_ratio=0.1 +# threshold_min_set=0.01 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done \ No newline at end of file diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..9c5a9c064af50f9f4813acc0b54a1f571cf709fd Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..e01bbe358016a399ce6db672338334c8df4584a0 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_gt.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..fdaf1345e0c5df06b42a0d5a9fd796c3af8a03a0 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/000/val/val_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/030/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..653f6af945db4b0734cb42c8812c013cbce1349b Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/030/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/060/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..e0480d848316636f87c33c2e016ee50e29fea272 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/EVA/renders_bg_360_3deg_35elev/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/060/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-023104.txt b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-023104.txt new file mode 100644 index 0000000000000000000000000000000000000000..089a22e6f57e937e71749ac1e76996808a03c623 --- /dev/null +++ b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-023104.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-044-041bc4c509634e52aa02d0bff5bf1ad8-000 --func_name main1 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-023302.txt b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-023302.txt new file mode 100644 index 0000000000000000000000000000000000000000..9d4ee2cd4385339e9ca3fb129313297b36c73752 --- /dev/null +++ b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/commandline_args2025-09-22-023302.txt @@ -0,0 +1 @@ +graph_model_preprocessing5.py --seq_name 000-044-041bc4c509634e52aa02d0bff5bf1ad8-000 --func_name main23 --dataset_name objaverse --depth_ratio 1.0 --threshold_min_set 0.01 --version_key_edges max_contrib --log_subfolder objaverse_fit_native_add3 \ No newline at end of file diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..20cde3ed21a3abe21896ee2f142ba212d1c46453 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/015/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..67d7110ef51249d17162a52b4ff1a6922387b481 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/030/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..12b2a34e0844a769109a1e1fa601e9e5ca4ff5e6 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/045/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..5054aad58cc30697172dcd015ffdbd8fa0d1c6a7 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/060/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..580c40cd450496f4bcf3db1c7f46c7fc27afb68e Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/075/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..129b3af287fe61b340ca833ae04ecf3583715e75 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/090/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..b6794b36a01a80933c3b3ca9061a848d399c9024 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/105/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..317e42986360ccd05f2f063136d74fae98e10695 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/120/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..49cd86f3108532ea5fc4845f4fd06e99aeff4fc0 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/135/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..eb48f76961c91f77f248f19016a5483a269da3af Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/150/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..4e5103c1b16790c4425d4d3f4e9f5dbb3873452d Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/165/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..9aed1563324591ea8d2205ac5dc104926f4b1b30 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/downsample2/180/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt new file mode 100644 index 0000000000000000000000000000000000000000..4258bbf2f4a049a37ee2460c18998d49a41ce73d --- /dev/null +++ b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/execution_times.txt @@ -0,0 +1,6 @@ +Function Execution Times +======================== + +photometric_reconstruct_from_pretrained: 1748.1462 seconds + +Total execution time: 1748.1462 seconds diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..ed39005a9ee8c555ba7b086a5e5eb8e95815aa2a Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/mono_135/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx new file mode 100644 index 0000000000000000000000000000000000000000..b885e1bc6fc64bfe2f44253c19058f3a6389bb22 Binary files /dev/null and b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/mono_180/val/val2_mask_square.xlsx differ diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt new file mode 100644 index 0000000000000000000000000000000000000000..80894764bda64f81d2920c769e19e201462566db --- /dev/null +++ b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/saved_ugraph_model/step_1599/fps_eval.txt @@ -0,0 +1 @@ +FPS: 174.0430012538 diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py new file mode 100644 index 0000000000000000000000000000000000000000..7016b5610ec6e7552d04c0af22d9b5b124beafb4 --- /dev/null +++ b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/graph_model_preprocessing5.py @@ -0,0 +1,2928 @@ +import sys +import os +sys.path.append(os.path.dirname("../lib_render")) +sys.path.append(os.path.dirname("../lib_mosca/gs_utils")) +sys.path.append(os.path.dirname("..")) + +import torch +import numpy as np +from scipy.spatial.transform import Rotation +# import matplotlib.pyplot as plt +import pickle +import random +import glob + +# import pypose as pp +from torch import nn + +from itertools import chain +from scipy.spatial import KDTree,Delaunay,distance_matrix +import open3d as o3d +from collections import defaultdict +import roma + +from mpl_toolkits.mplot3d import Axes3D + +# import hdbscan +from sklearn.cluster import DBSCAN +from tqdm import tqdm + +from lib_mosca.mosca import __DQB_warp2__, __LBS_warp2__ + +from lib_mosca.gs_utils.utils_helper import project_points, draw_projected_points, draw_projected_points_masked, valid_and_visible, overlay_mask_yellow, stack_images_side_by_side, compute_depth_gradient, associate_pts_value, draw_projected_points_colored +from lib_mosca.gs_utils.som_vis_utils import ( + apply_depth_colormap, +) + +import matplotlib.pyplot as plt +import argparse +import logging, glob, sys, os, shutil, os.path as osp + +def get_order(arr): + """ + Computes the rank (order) of each element in the array. + + Args: + arr (np.ndarray): Input 1D array. + + Returns: + np.ndarray: Array where the i-th element is the rank of the i-th element in the input array. + """ + # Get the indices that would sort the array + sorted_indices = np.argsort(arr) + + # Create an array to store the ranks + order = np.empty_like(sorted_indices) + + # Assign ranks + order[sorted_indices] = np.arange(len(arr)) + + return order + +def count_out_dict_files(folder_path): + # Use glob to find files matching the pattern + file_pattern = os.path.join(folder_path, 'out_dict_*.pkl') + files = glob.glob(file_pattern) + return len(files) + +def calculate_edge_relative_orientation(Ts): # 854 pts will use ?? GB memory!!! + nk=Ts.shape[1] + Ts1=Ts.unsqueeze(2).repeat(1,1,nk,1) + Ts2=Ts.unsqueeze(1).repeat(1,nk,1,1) + Ts12=(Ts1.Inv()*Ts2) + # Ts_rel=relativelieAlgebra_pp(Ts1,Ts2) + return Ts12 + +def get_navie_ori_distances(Oris_rel): # TODO + Oris_rel_so3=Oris_rel.Log() # (nt, nk, nk, 3) + norm_oris_rel_so3=torch.norm(Oris_rel_so3.tensor(),dim=-1) # (nt, nk, nk) + norm_ori_rel_so3=torch.mean(norm_oris_rel_so3,axis=0) # (nk, nk) + return 1.0 - torch.cos(norm_ori_rel_so3) # (nk, nk) + + +def compute_accel_loss(transls): + # transls [nt,n,3] or oris [nt,n,4] + accel = 2 * transls[1:-1] - transls[:-2] - transls[2:] # [nt-2,n,3/4] + loss = accel.norm(dim=-1).mean() # [nt-2,n,3/4] => [nt-2,n]=> [1] + return loss +def cal_rigidity_loss(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global + stats = { + "loss_local_rigidity": loss_local_rigidity, + "loss_local_rigidity_ori": loss_local_rigidity_ori, + "loss_local_global": loss_local_global, + } + + return loss, stats + +def cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges): + oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + return loss_local_rigidity_dt,loss_local_rigidity_ori_dt + +def cal_rigidity_loss_extra(transls,oris_xyzw,edges_local,weight_edges,dis_local_canonical_edge): + # transls.shape # [nt, num_nodes, 3] + # oris_xyzw.shape # [nt, num_nodes, 4] + # edges_local.shape # [num_edges, 2] # [...[center id, neighbor id]...] + # weight_edges.shape # [num_edges] + # dis_local_canonical_edge.shape # [num_edges] + + nt=transls.shape[0] + + # 1. compute local rigidity loss for transls + # weight_edges=weights.reshape(-1) # [num_edges] + transls_diff_edges=transls[:,edges_local[:,1]]-transls[:,edges_local[:,0]] # [nt, num_edges, 3] + # oris_xyzw_i=oris_xyzw[:,edges_local[:,0]] # [:,num_edges, 4] + + # t, t-1 + oris_xyzw_tdiff=roma.quat_product(oris_xyzw[:-1],roma.quat_conjugation(oris_xyzw[1:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + oris_xyzw_tdiff_edges=oris_xyzw_tdiff[:,edges_local[:,0]] # [nt-1, num_edges, 4] + transls_diff_cal_edges=roma.utils.quat_action(oris_xyzw_tdiff_edges,transls_diff_edges[1:]) # [nt-1, num_edges, 3] # = R*t + transls_diff_current=transls_diff_edges[:-1] # [nt-1, num_edges, 3] + dis_diff=torch.norm(transls_diff_cal_edges-transls_diff_current,dim=-1) # [nt-1, num_edges] + error_local_rigidity=dis_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity=error_local_rigidity.mean() # scalar + + # 2. compute local rigidity loss for oris + qq_j_edges=oris_xyzw_tdiff[:,edges_local[:,1]] # [nt-1, num_edges, 4] + qq_i_edges=oris_xyzw_tdiff[:,edges_local[:,0]] + qq_diff=torch.norm(qq_j_edges-qq_i_edges,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + error_local_rigidity_ori=qq_diff*weight_edges[None,:] # [nt-1, num_edges] + loss_local_rigidity_ori=error_local_rigidity_ori.mean() # scalar + + # 3. compute global rigidity loss + # dis_local_canonical=sq_dists_local**0.5 # [num_nodes, n_neighbors] + # dis_local_canonical_edge=dis_local_canonical.reshape(-1) # [num_edges] # TODO: CHECK + dis_local_edge=torch.norm(transls_diff_edges,dim=-1) # [nt, num_edges] + dis_local_diff=torch.abs(dis_local_canonical_edge[None,:]-dis_local_edge) # [nt, num_edge] + error_local_global=dis_local_diff*weight_edges[None,:] # [nt, num_edge] + loss_local_global=error_local_global.mean() # scalar + + # 4. compute dt rigidity loss + # delta_t=50 + # oris_xyzw_tdiff_dt=roma.quat_product(oris_xyzw[:-delta_t],roma.quat_conjugation(oris_xyzw[delta_t:])) # [nt-1, num_nodes, 4] # R{i,t-1} * R{i,t}^-1 # Conjugation of a unit quaternion is equal to its inverse. # quat_inverse(quat) + # oris_xyzw_tdiff_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] # [nt-1, num_edges, 4] + # transls_diff_cal_edges_dt=roma.utils.quat_action(oris_xyzw_tdiff_edges_dt,transls_diff_edges[delta_t:]) # [nt-1, num_edges, 3] # = R*t + # transls_diff_current_dt=transls_diff_edges[:-delta_t] # [nt-1, num_edges, 3] + # dis_diff_dt=torch.norm(transls_diff_cal_edges_dt-transls_diff_current_dt,dim=-1) # [nt-1, num_edges] + # error_local_rigidity_dt=dis_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_dt=error_local_rigidity_dt.mean() # scalar + + # # 2. compute local rigidity loss for oris + # qq_j_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,1]] # [nt-1, num_edges, 4] + # qq_i_edges_dt=oris_xyzw_tdiff_dt[:,edges_local[:,0]] + # qq_diff_dt=torch.norm(qq_j_edges_dt-qq_i_edges_dt,dim=-1) # [nt-1, num_edges] ### TODO: can change to SO3 version + # error_local_rigidity_ori_dt=qq_diff_dt*weight_edges[None,:] # [nt-1, num_edges] + # loss_local_rigidity_ori_dt=error_local_rigidity_ori_dt.mean() # scalar + + if nt<100: + delta_t=int(nt/2) + else: + delta_t=50 + loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(delta_t,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + # loss_local_rigidity_50,loss_local_rigidity_ori_50=cal_dt_loss(50,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss_local_rigidity_10,loss_local_rigidity_ori_10=cal_dt_loss(10,oris_xyzw,edges_local,weight_edges,transls_diff_edges) + + loss = loss_local_rigidity+loss_local_rigidity_ori+100*loss_local_global+100*loss_local_rigidity_50+100*loss_local_rigidity_10#+100*loss_local_rigidity_ori_dt + stats = { + "loss_local_rigidity": loss_local_rigidity.item(), + "loss_local_rigidity_ori": loss_local_rigidity_ori.item(), + "loss_local_global": loss_local_global.item(), + "loss_local_rigidity_50": loss_local_rigidity_50.item(), + "loss_local_rigidity_10": loss_local_rigidity_10.item(), + } + + return loss, stats + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +def generate_ratio_binary_tensor(n, ratio_0=0.66): + # assert abs(ratio_0 + ratio_1 - 1.0) < 1e-6, "Ratios must sum to 1" + ratio_1=1-ratio_0 + + num_0 = int(round(n * ratio_0)) + num_1 = n - num_0 # to ensure total length is exactly n + + tensor = torch.cat([ + torch.zeros(num_0, dtype=torch.bool), + torch.ones(num_1, dtype=torch.bool) + ]) + + shuffled_tensor = tensor[torch.randperm(n)] + return shuffled_tensor + +# optimize_overall_with_orientation version of +def optimize_overall_with_orientation_edges( + edges, + Transls, Transls_optimizable, Mask, Dists_optimizable, weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, Ori_distances_optimizable, weight_ori_distance_optimizable, + Dists_observed_mean, Dists_observed_std, Mask_observed,Nearby_mask, + W2Cs,Depths_info_perG_key, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 ###### test + + # Depths_info_perG_key [nt,nk] + + nt = Transls.shape[0] + nk = Transls.shape[1] + ne = edges.shape[0] + + # Device setup + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + Dists_optimizable = nn.Parameter(Dists_optimizable.to(device)) + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Ori_distances_optimizable = nn.Parameter(Ori_distances_optimizable.to(device)) + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Mask = Mask.to(device) + Depths_info_perG_key=Depths_info_perG_key.to(device) + + optimizer = torch.optim.Adam( + [Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable, Ori_distances_optimizable], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, threshold=1e-3, min_lr=min_lr + ) + + # Distance matrix helper function + def distance_matrix(x, y): + return torch.cdist(x, y, p=2) + + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt,3,3] + # I0=torch.eye(3,device=device) + # I0[2,2]=0.1 # 0.01 camel + # I0=I0[None,:,:] + # I1=W2Cs_R.transpose(-1, -2)@I0@W2Cs_R # [nt,3,3] + # I1=I1[:,None,:,:] + I0=torch.eye(3,device=device).expand(nt,nk,-1,-1).clone() # [nt,nk,3,3] + I0[...,-1,-1]=Depths_info_perG_key # 0.01 camel # CHECK TODO + I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nk,3,3] + # breakpoint() + + # Optimization loop + # for epoch in range(max_epochs): + for epoch in tqdm(range(max_epochs), desc="Training Epochs", unit="epoch"): + optimizer.zero_grad() + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + a1=(Transls_optimizable-Transls)[:,:,None,:] # (nt, nk, 1, 3) + a1_T=a1.transpose(-1, -2) + squared_norm=(a1@I1@a1_T).squeeze(-1,-2) + # squared_norm=(a1@a1_T).squeeze(-1,-2) # test + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # add epsilon for numerical stability + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) # full model + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) # ablation of loss + # Dist_confident_pts_error_masked=Dist_confident_pts_error[Mask] + # Dist_confident_pts_error_masked_disturbed=Dist_confident_pts_error_masked+torch.randn_like(Dist_confident_pts_error_masked)*0.01 + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed) + # mask_shuffle=generate_ratio_binary_tensor(n=Dist_confident_pts_error_masked_disturbed.shape[0], ratio_0=0.8) + # mask_shuffle=torch.ones_like(Dist_confident_pts_error_masked_disturbed,dtype=torch.bool) + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error_masked_disturbed[mask_shuffle]) + + + # Ori_confident_pts_error=torch.norm(Oris_xyzw_optimizable_normed-Oris_xyzw,dim=-1) # test + # loss_oris_confident_pts = torch.mean(Ori_confident_pts_error[Mask]) + + mask_edge=generate_ratio_binary_tensor(n=edges.shape[0], ratio_0=0.3) + edges_mask=edges[mask_edge] + weight_normed=torch.abs(weight_distance_optimizable)/torch.norm(weight_distance_optimizable,dim=-1,p=1)[:,None] # (nu,nk) + weight_edges=weight_normed[edges_mask[:,0],edges_mask[:,1]] # (ne) + dis_local_canonical_edge=Dists_optimizable[edges_mask[:,0],edges_mask[:,1]] # (ne) CHECK order + # loss_rigidity,stats_rigidity=cal_rigidity_loss(Transls_optimizable,Oris_xyzw_optimizable_normed,edges,weight_edges,dis_local_canonical_edge) + loss_rigidity,stats_rigidity=cal_rigidity_loss_extra(Transls_optimizable,Oris_xyzw_optimizable_normed,edges_mask,weight_edges,dis_local_canonical_edge) + + # Loss 3: Regularization on weight distance + loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch[:-1] - Dists_epoch[1:])) # (nt-1,nk,nk) + # Dists_epoch_edges2=Dists_epoch_edges.reshape(nt,ne) # (nt,ne,nk) + # loss_local_smoothness = torch.mean(torch.abs(Dists_epoch_edges2[:-1] - Dists_epoch_edges2[1:])) # (nt-1,nk,nk) + # loss_local_smoothness = torch.mean(torch.abs(Transls_optimizable[:-1]-Transls_optimizable[1:])) # (nt-1,nk,nk) # [TODO] + loss_accel = compute_accel_loss(Transls_optimizable) + + + # Loss 7: symmetry loss for edge weights + # TODO + # loss_symmetry=torch.mean(torch.abs(weight_distance_optimizable-weight_distance_optimizable.transpose(0,1))) + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + # a_loss_oris_confident_pts=1*loss_oris_confident_pts + a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point 0.1 common value + a_loss_sparsity=1e-1/loss_sparsity + a_loss_accel=0.1*loss_accel + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # flag_train_orientation=True + # if flag_train_orientation: + # a_loss_density=5e-3/loss_sparsity # 1e-3 for pos+rotation + # else: + # a_loss_density=5e-4/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + # a_loss_oris_distance_confident_pts=1*loss_oris_distance_confident_pts + # a_loss_ori_dist=1*loss_ori_dist + # a_loss_ori_local_smoothness=0.01*loss_ori_local_smoothness + + total_loss = a_loss_dist_confident_pts + a_loss_rigidity + a_loss_accel # + a_loss_sparsity + # + a_loss_dist + a_loss_local_smoothness + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness + a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # + a_loss_oris_confident_pts + + # Backpropagation and optimization step + total_loss.backward() + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # Logging progress + rigidity1=stats_rigidity["loss_local_rigidity"] + rigidity_ori2=stats_rigidity["loss_local_rigidity_ori"] + global3=stats_rigidity["loss_local_global"] + if epoch % log_interval == 0 or epoch == max_epochs - 1 or optimizer.param_groups[0]['lr'] < min_lr * 2: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + # f"a_loss_density: {a_loss_density.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + # f"OConfDist: {a_loss_oris_confident_pts.item():.6f}, " + f"a_ridigity: {a_loss_rigidity.item():.6f}, " + # f"rigidity1:{rigidity1:.6f}, " + # f"rigidity_ori2:{rigidity_ori2:.6f}, " + # f"global3:{global3:.6f}, " + + # f"Dist: {a_loss_dist.item():.6f}, " + # f"OConfDist {a_loss_oris_distance_confident_pts.item():.6f}, " + # f"ODist {a_loss_ori_dist.item():.6f}, " + # f"Smooth: {a_loss_local_smoothness.item():.6f}, " + # f"OSmooth: {a_loss_ori_local_smoothness.item():.6f}, " + # f"Rigidity L: {loss_local_rigidity_2.item():.6f}, " + f"a_loss_accel: {a_loss_accel.item():.6f}, " + f"a_loss_sparsity: {a_loss_sparsity.item():.6f}, " + # f"Symmetry L: {a_loss_symmetry.item():.6f}, " + # f"W L: {loss_weight_distance.item():.6f}, " + # f"W L2: {loss_weight_distance_2.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + print(f"weight_normed: {(weight_normed>0.1).sum().item():.6f}") + # print(stats_rigidity) + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and epoch > 5000: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + return Transls_optimizable, Dists_optimizable, weight_distance_optimizable, Oris_xyzw_optimizable_normed, Ori_distances_optimizable + + +def get_knn_edges(n_neighbors,Transls,Contribs,Transls2,Contribs2,contrib_threshold, flag_remove_self=False): + # choice 2: knn graph + def o3d_knn_p(pts2,pts, num_knn, flag_remove_self=False): + indices = [] + sq_dists = [] + pcd = o3d.geometry.PointCloud() + pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + pcd_tree = o3d.geometry.KDTreeFlann(pcd) + for p in pts2: + if not flag_remove_self: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + indices.append(i[0:]) + sq_dists.append(d[0:]) + else: + [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn + 1) + indices.append(i[1:]) + sq_dists.append(d[1:]) + return np.array(sq_dists), np.array(indices) + + nk=Transls.shape[1] + nu=Transls2.shape[1] + + i_s_knn_others,i_s_knn_is=[],[] + # n_neighbors=5 + for i in range(nu): + t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + pts2=[pt2] # just one point + sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors, flag_remove_self) + sq_dists,indices= sq_dists[0],indices[0] + i_s_knn=i_s[indices] # usage: ids[i_s_knn] + i_s_knn_others.append(i_s_knn) + i_s_knn_is.append(i*torch.ones_like(i_s_knn)) + i_s_knn_is=torch.cat(i_s_knn_is) + i_s_knn_others=torch.cat(i_s_knn_others) + edges_others=torch.stack([i_s_knn_is,i_s_knn_others],dim=-1) + # breakpoint() + # old version: has bug when knn key points are not enough + # i_s_knn_others=torch.stack(i_s_knn_others) + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # edges_others.shape # nu*5 + return edges_others + +# def get_knn_edges_naive(n_neighbors,Transls,Transls2, flag_remove_self=False): +# nk=Transls.shape[1] +# nu=Transls2.shape[1] +# nt=Transls.shape[0] +# Dist=[] +# for t in range(nt): +# dist=(Transls2[t][:,None]-Transls[t][None]).norm(dim=-1) +# Dist.append(dist) +# Dist=torch.stack(Dist,dim=0) +# Dist=Dist.mean(dim=0) + +# def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): +# """ +# Naive temporal k-NN graph construction using average distance over time. + +# Args: +# n_neighbors (int): Number of neighbors. +# Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. +# Transls2 (torch.Tensor): (T, N2, D) tensor of query points. +# flag_remove_self (bool): If True and N1 == N2, removes self-matches. + +# Returns: +# torch.Tensor: (N2 * n_neighbors, 2) edge index pairs [Transls2_index, Transls_index]. +# """ +# T, N1, D = Transls.shape +# N2 = Transls2.shape[1] + +# # Compute temporal average pairwise distances +# dist_list = [] +# for t in range(T): +# dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) +# dist_list.append(dist) +# Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + +# if flag_remove_self and N1 == N2: +# diag = torch.arange(N2, device=Dist.device) +# Dist[diag, diag] = float('inf') + +# # Get top-k nearest neighbors for each Transls2 point +# _, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + +# # Build edge index pairs +# row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) +# edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + +# return edges_others + +def get_knn_edges_naive(n_neighbors, Transls, Transls2, flag_remove_self=False): + """ + Naive temporal k-NN graph construction using average distance over time. + + Args: + n_neighbors (int): Number of neighbors. + Transls (torch.Tensor): (T, N1, D) tensor of reference/key points. + Transls2 (torch.Tensor): (T, N2, D) tensor of query points. + flag_remove_self (bool): If True and N1 == N2, removes self-matches. + + Returns: + edges_others (torch.Tensor): (N2 * k, 2) edge index pairs [Transls2_index, Transls_index]. + dists (torch.Tensor): (N2 * k,) mean distances corresponding to each edge. + """ + T, N1, D = Transls.shape + N2 = Transls2.shape[1] + + # Compute temporal average pairwise distances + dist_list = [] + for t in range(T): + dist = (Transls2[t][:, None, :] - Transls[t][None, :, :]).norm(dim=-1) # (N2, N1) + dist_list.append(dist) + Dist = torch.stack(dist_list, dim=0).mean(dim=0) # (N2, N1) + + if flag_remove_self and N1 == N2: + diag = torch.arange(N2, device=Dist.device) + Dist[diag, diag] = float('inf') + + # Get top-k nearest neighbors and distances + dists, knn_indices = torch.topk(Dist, k=n_neighbors, dim=1, largest=False) # (N2, k) + + # Build edge index pairs + row_indices = torch.arange(N2, device=Dist.device).unsqueeze(1).repeat(1, n_neighbors) # (N2, k) + edges_others = torch.stack([row_indices, knn_indices], dim=-1).reshape(-1, 2) # (N2 * k, 2) + dists_edges = dists.reshape(-1) # (N2 * k,) + + return edges_others, dists_edges + + + + + +def optimize_others_with_orientation_edges4( + edges, + Transls, Transls_optimizable, Mask, #Dists_optimizable, #weight_distance_optimizable, + Oris_xyzw, Oris_xyzw_optimizable, #Ori_distances_optimizable, #weight_ori_distance_optimizable, + Transls_key, Oris_xyzw_key,Transls_key_optimized, Oris_xyzw_key_optimized, + weight_param_key_optimizable,sq_dists_key_nonkey_edges, + ts_most_confident,per_nonkeyGaussian_dweight, + W2Cs,Depths_info_perG_nonkey, + device, + lr=1e-1, max_epochs=10000, min_lr=1e-6, log_interval=100 +): + # test + # max_epochs=10 + # weight_param_key_optimizable [nk] per-keyGuassian weight + # Depths_info_perG_nonkey [nt,nu] + # nt = ts.shape[0] + + + nt_all = Transls.shape[0] + nu=Transls.shape[1] + nk=Transls_key_optimized.shape[1] + ne=edges.shape[0] + + Oris_xyzw=Oris_xyzw/Oris_xyzw.norm(dim=-1,keepdim=True) # normalize + # Oris_xyzw_optimizable=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) # normalize + Oris_xyzw_key_optimized=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) # normalize + + # Device setup + # Transls_optimizable = nn.Parameter(Transls_optimizable.to(device)) + # Oris_xyzw_optimizable = nn.Parameter(Oris_xyzw_optimizable.tensor().to(device)) + Transls_optimizable=Transls_optimizable.to(device) + Oris_xyzw_optimizable=Oris_xyzw_optimizable.tensor().to(device) + + Transls = Transls.to(device) + Oris_xyzw = Oris_xyzw.to(device) + Transls_key = Transls_key.to(device) + Oris_xyzw_key = Oris_xyzw_key.to(device) + Transls_key_optimized = Transls_key_optimized.to(device).detach() + Oris_xyzw_key_optimized = Oris_xyzw_key_optimized.to(device).detach() + weight_param_key_optimizable = nn.Parameter(weight_param_key_optimizable.to(device)) + # sq_dists_key_nonkey = sq_dists_key_nonkey.to(device) + sq_dists_key_nonkey_edges = sq_dists_key_nonkey_edges.to(device) + per_nonkeyGaussian_dweight = nn.Parameter(per_nonkeyGaussian_dweight.to(device)) # don't use for now, just keep + # per_nonkeyGaussian_dweight=None + Depths_info_perG_nonkey = Depths_info_perG_nonkey.to(device) + + # sq_dists_key_nonkey_edges=sq_dists_key_nonkey[edges[:,0],edges[:,1]] # (ne) + exp_sq_dists_key_nonkey_edges=torch.exp(-sq_dists_key_nonkey_edges) # (ne) + W2Cs_R=torch.tensor(W2Cs[:,:3,:3],dtype=torch.float32,device=device) # [nt_all,3,3] + + Mask = Mask.to(device) + + # Distance matrix helper function + # def distance_matrix(x, y): + # return torch.cdist(x, y, p=2) + + @torch.no_grad() # check + def cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu): + # weight_param_key_optimizable [nk] + # exp_sq_dists_key_nonkey_edges [ne] + device=weight_param_key_optimizable.device + nk=weight_param_key_optimizable.shape[0] + + weight_param_key_optimizable_k=100*torch.abs(weight_param_key_optimizable) # [TODO]: 2000 # 100 + weight_param_edges=weight_param_key_optimizable_k[edges[:,1]] # (ne) + # assert all(weight_param_edges>0) + # assert all(sq_dists_key_nonkey_edges>0) + # weight_edges = (exp_sq_dists_key_nonkey_edges**(1/10))**weight_param_edges # TODO: to unify scale #test: 10 for block + weight_edges = exp_sq_dists_key_nonkey_edges**weight_param_edges # TODO: to unify scale #test: 20 for block + + weight_edges = torch.clamp(weight_edges, min=1e-10, max=1e10) # test + + weights_matrix=torch.zeros((nu,nk),device=device) + weights_matrix[edges[:,0],edges[:,1]]=weight_edges + sum_temp = torch.sum(weights_matrix,dim=-1,keepdim=True) + if any(sum_temp==0): + breakpoint() + if torch.isnan(weight_param_key_optimizable).any(): + breakpoint() + weights_matrix_normed=weights_matrix/sum_temp + + weight_edges_normed=weights_matrix_normed[edges[:,0],edges[:,1]] # (ne) + + return weight_edges_normed + + + + # prepare + Oris_xyzw_key_norm=Oris_xyzw_key/Oris_xyzw_key.norm(dim=-1,keepdim=True) + Oris_xyzw_key_optimized_norm=Oris_xyzw_key_optimized/Oris_xyzw_key_optimized.norm(dim=-1,keepdim=True) + n_neighbors=(edges[:,0]==0).sum() + if n_neighbors!=5: + breakpoint() + + # new + src_xyz_1=Transls[ts_most_confident,edges[::n_neighbors,0]] # (nu,3) + src_xyzw_1=Oris_xyzw[ts_most_confident,edges[::n_neighbors,0]] # (nu,4) + src_xyz_1 = nn.Parameter(src_xyz_1) + src_xyzw_1 = nn.Parameter(src_xyzw_1) + + # get sk_src_node and sk_dst_node (key nodes) + ts_most_confident_repeat=ts_most_confident.repeat(n_neighbors,1).T.reshape(-1) # (nt*5) + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + # optimizer = torch.optim.Adam( + # [weight_param_key_optimizable,src_xyz_1,src_xyzw_1,per_nonkeyGaussian_dweight], lr=lr + # ) + optimizer = torch.optim.Adam( + [weight_param_key_optimizable,src_xyz_1,src_xyzw_1], lr=lr + ) + + scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau( + optimizer, mode='min', factor=0.5, patience=10, min_lr=min_lr + ) + ######################################################3 + + # ts_batches = torch.split(ts_all, batch_size) + # for ts_batch in ts_batches: + # Optimization loop + # for epoch in range(max_epochs): + # ts_all=torch.arange(nt_all,device=device) + # batch_size=30 + # pbar=tqdm(range(max_epochs)) + # for step in pbar: + # optimizer.zero_grad() + ts_all = torch.arange(nt_all, device=device) + batch_size = 30 + steps_per_epoch = (nt_all + batch_size - 1) // batch_size + total_steps = steps_per_epoch * max_epochs + + pbar = tqdm(range(total_steps), desc="Training") + + for step in pbar: + epoch = step // steps_per_epoch + inner_step = step % steps_per_epoch + + # Shuffle once per epoch + if inner_step == 0: + ts_shuffled = ts_all[torch.randperm(nt_all)] + + # Select current batch + start = inner_step * batch_size + ts_batch = ts_shuffled[start:start + batch_size] + pbar.set_description(f"Epoch {epoch}, Step {inner_step}/{steps_per_epoch-1}") + + optimizer.zero_grad() + + # print(f"ts_batch: {ts_batch}") + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # Mask_DQB=Mask[ts_batch] + + sk_nt_node_xyz_1=Transls_key[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_2=sk_nt_node_xyz_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_1=Oris_xyzw_key_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_1=sk_nt_node_xyzw_1[:,:,[3,0,1,2]] + sk_nt_node_quat_2=sk_nt_node_wxyz_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt, -1, -1, -1)[Mask_DQB] # (n_mask,n_neighbors,4) + sk_nt_dst_node_xyz=sk_nt_node_xyz_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat=sk_nt_node_quat_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # error propagation + I0=torch.eye(3,device=device).expand(nt,nu,-1,-1).clone() # [nt,nu,3,3] + # I0[...,-1,-1]=Depths_info_perG_nonkey # 0.001 camel # CHECK TODO + # I1=W2Cs_R.transpose(-1, -2)[:,None,:,:]@I0@W2Cs_R[:,None,:,:] # [nt,nu,3,3] + I1=I0 # test # [nt,nu,3,3] + I1_masked=I1[Mask_DQB] # [n_mask,3,3] + + if torch.isnan(weight_param_key_optimizable).any(): + print("weight_param_key_optimizable is nan") + breakpoint() + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + src_xyz=src_xyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt, -1, -1)[Mask_DQB] # (n_mask<=nt*nu,4) + + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) # [nu,n_neighbors] + # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # sk_w=weight_per_nonkeyGaussian_adjusted.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz, # [N,K,3] + sk_nt_dst_node_quat, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + # Transls_interpolation=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + # Oris_interpolation=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_interpolation_masked=mu_dst # [n_mask,3] + Oris_interpolation_masked=quat_dst_xyzw # [n_mask,4] + + + # error propagation + # a1=(Transls_interpolation[Mask]-Transls[Mask])[:,None,:] # (n_mask, 1, 3) + a1=(Transls_interpolation_masked-Transls[ts_batch][Mask_DQB])[:,None,:] # (n_mask, 1, 3) + # a1=(Transls_interpolation_masked-Transls[ts_batch])[:,None,:] # (n_mask, 1, 3) # ablation of loss + a1_T=a1.transpose(-1, -2) # (n_mask, 3, 1) + squared_norm=(a1@I1_masked@a1_T).squeeze(-1,-2) # [n_mask] + Dist_confident_pts_error = torch.sqrt(squared_norm + 1e-8) # [n_mask] # add epsilon for numerical stability + loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Loss 1: Confident points distance error + # Dist_confident_pts_error=torch.norm(Transls_optimizable-Transls,dim=-1) # [nt,nu] + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error[Mask]) + # Dist_confident_pts_error=torch.norm(Transls_interpolation[Mask]-Transls[Mask],dim=-1) # [nt,nu] # last one + # loss_dist_confident_pts = torch.mean(Dist_confident_pts_error) + + # Ori_confident_pts_error=torch.norm(Oris_interpolation[Mask]-Oris_xyzw[Mask],dim=-1) # [n_mask] + Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] + # Ori_confident_pts_error=torch.norm(Oris_interpolation_masked-Oris_xyzw[ts_batch][Mask_DQB],dim=-1) # [n_mask] # ablation of loss + loss_ori_confident_pts = torch.mean(Ori_confident_pts_error) + + # Loss 3: Regularization on weight distance + # loss_sparsity=torch.mean(torch.norm(weight_normed,p=1,dim=-1)/torch.norm(weight_normed, p=2,dim=-1) )-1 # float('inf') + + # Loss 5: local time smoothness (help reduce deviation loss) + # loss_velocity = torch.mean(torch.abs(Transls_interpolation[:-1]-Transls_interpolation[1:])) # (nt-1,nk,nk) # [TODO] + # loss_ori_velocity = torch.mean(torch.abs(Oris_interpolation[:-1] - Oris_interpolation[1:])) + + # loss_accel = compute_accel_loss(Transls_interpolation) + # loss_ori_accel=compute_accel_loss(Oris_interpolation) + + + # Loss 6: regularize per nonkey Gaussian delta weight + # loss_per_nonkeyGaussian=torch.mean(torch.abs(per_nonkeyGaussian_dweight)) + # loss_per_nonkeyGaussian = 1000*torch.mean(per_nonkeyGaussian_dweight ** 2) # TODO + + # Adding other loss may make weights too small.=> point prediction by graph does not work well. + # [To solve the problem, regularize W to avoid it too small.] + # Total loss + a_loss_dist_confident_pts=1*loss_dist_confident_pts + a_loss_ori_confident_pts=1*loss_ori_confident_pts + # a_loss_interpolation=1*loss_interpolation # 100 + # a_loss_interpolation_ori=1*loss_interpolation_ori # 100 + # a_loss_rigidity=1*loss_rigidity + # a_loss_dist=100*loss_dist + # a_loss_local_smoothness=0.1*loss_local_smoothness # 0.1 has jumping point + + # a_loss_velocity=0.1*loss_velocity + # a_loss_ori_velocity=0.1*loss_ori_velocity + # a_loss_accel=0.1*loss_accel + # a_loss_ori_accel=0.1*loss_ori_accel + + # a_loss_per_nonkeyGaussian=1e-2*loss_per_nonkeyGaussian + + # a_loss_sparsity=1e-3*loss_sparsity + + # a_loss_density=5e-2/loss_sparsity # 1e-4 for pos only + # a_loss_distance_deviation=1*loss_distance_deviation + # a_loss_symmetry=1e-4*loss_symmetry + + total_loss = a_loss_dist_confident_pts + a_loss_ori_confident_pts #+ a_loss_per_nonkeyGaussian#+ \ + # a_loss_velocity + a_loss_ori_velocity + a_loss_accel + a_loss_ori_accel #+ a_loss_sparsity \ + # a_loss_dist + # + a_loss_oris_distance_confident_pts + a_loss_ori_dist + a_loss_ori_local_smoothness#+0.1*loss_local_rigidity_2+0.1*loss_weight_distance_2#+0.001*loss_sparsity#+ 0.1*loss_distance_deviation #+ 0.1*loss_weight_distance+ 0.1*loss_weight_distance_2 + #+ a_loss_local_smoothness+ a_loss_sparsity + a_loss_symmetry\ + # + a_loss_density + # a_loss_rigidity + # a_loss_interpolation + a_loss_interpolation_ori + \ + + # Backpropagation and optimization step + total_loss.backward() + # total_loss.backward(retain_graph=True) + optimizer.step() + + # Adjust learning rate with the scheduler + scheduler.step(total_loss) + + # # Logging progress + if epoch % log_interval == 0 or epoch == max_epochs - 1: + print(f"Epoch {epoch}, L: {total_loss.item():.6f}, " + f"ConfDist: {a_loss_dist_confident_pts.item():.6f}, " + f"a_loss_ori_confident_pts: {a_loss_ori_confident_pts.item():.6f}, " + # f"a_loss_per_nonkeyGaussian: {a_loss_per_nonkeyGaussian.item():.6f}, " + # f"a_loss_interpolation: {a_loss_interpolation.item():.6f}, " + # f"a_loss_interpolation_ori: {a_loss_interpolation_ori.item():.6f}, " + # f"a_loss_rigidity: {a_loss_rigidity.item():.6f}, " + # f"a_loss_velocity: {a_loss_velocity.item():.6f}, " + # f"a_loss_ori_velocity: {a_loss_ori_velocity.item():.6f}, " + # f"a_loss_accel: {a_loss_accel.item():.6f}, " + # f"a_loss_ori_accel: {a_loss_ori_accel.item():.6f}, " + # f"Density L: {a_loss_density.item():.6f}, " + f"LR: {optimizer.param_groups[0]['lr']:.2e}") + # # print(f"weight_normed: {(weight_normed>1e-6).sum().item():.6f}") + # # print(f"weight_param_key_optimizable:" {weight_param_key_optimizable}) + + pbar.set_description(f"L:{total_loss.item():.6f}, "+ + f"c:{a_loss_dist_confident_pts.item():.6f}, "+ + f"oc:{a_loss_ori_confident_pts.item():.6f}, "+ + # f"per:{a_loss_per_nonkeyGaussian.item():.6f}, "+ + # f"i: {a_loss_interpolation.item():.6f}, "+ + # f"io: {a_loss_interpolation_ori.item():.6f}, "+ + # f"r:{a_loss_rigidity.item():.6f}, "+ + # f"b:{a_loss_velocity.item():.6f}, "+ + # f"ov:{a_loss_ori_velocity.item():.6f}, "+ + # f"a:{a_loss_accel.item():.6f}, "+ + # f"oa:{a_loss_ori_accel.item():.6f}, "+ + f"lr:{optimizer.param_groups[0]['lr']:.2e}") + + # Early stopping + if optimizer.param_groups[0]['lr'] < min_lr * 2: #and a_loss_dist < 1e-3 and a_loss_oris_distance_confident_pts < 1e-1: + print(f"Stopping early at epoch {epoch}.") + break + + # Oris_xyzw_optimizable_normed=Oris_xyzw_optimizable/Oris_xyzw_optimizable.norm(dim=-1,keepdim=True) + + # # TEMP: for keeping data structure + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + # sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + # # weight_per_nonkeyGaussian_adjusted=torch.abs(weight_per_nonkeyGaussian+per_nonkeyGaussian_dweight) + # # weight_per_nonkeyGaussian_adjusted=weight_per_nonkeyGaussian_adjusted/torch.sum(weight_per_nonkeyGaussian_adjusted,dim=-1,keepdim=True) # (nu,n_neighbors) + # # sk_w=weight_per_nonkeyGaussian_adjusted[None,:,:].repeat(nt,1,1)[Mask_DQB] # (n_mask,n_neighbors) + + # weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + # weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + # # for i,e in enumerate(edges): + # # weight_distance_optimizable[e[0],e[1]]=weight_edges[i] + # weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + # print("End of optimize_others_with_orientation_edges4.") + # # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + # print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + # blend_dict = { + # 'sk_w': sk_w, + # 'src_xyz': src_xyz, + # 'src_quat': src_quat, + # 'sk_nt_src_node_xyz': sk_nt_src_node_xyz, + # 'sk_nt_src_node_quat': sk_nt_src_node_quat, + # 'sk_nt_dst_node_xyz': sk_nt_dst_node_xyz, + # 'sk_nt_dst_node_quat': sk_nt_dst_node_quat + # } + + + + # ********************* new trying begin ******************* + # Step 1: calculate src_xyz_1_new and src_xyzw_1_new + weight_param_key_optimizable=weight_param_key_optimizable.abs() + weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) # (nu,n_neighbors) + + weight_distance_optimizable= torch.zeros((nu,nk),dtype=torch.float32,device=device) # TODO + weight_distance_optimizable[edges[:,0],edges[:,1]]=weight_edges + weight_distance_optimizable = nn.Parameter(weight_distance_optimizable.to(device)) + + # print(np.histogram((weight_per_nonkeyGaussian_adjusted>0.1).sum(dim=-1).cpu().detach().numpy(),bins=4)) + print("hist weight_edges:",np.histogram(weight_edges.cpu().detach().numpy(),bins=5)) + + sk_w=weight_per_nonkeyGaussian # (nu,n_neighbors) + + sk_dst_node_xyz_1=Transls_key_optimized[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_dst_node_xyzw_1=Oris_xyzw_key_optimized_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_dst_node_wxyz_1=sk_dst_node_xyzw_1[:,[3,0,1,2]] + sk_dst_node_xyz_2=sk_dst_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_dst_node_quat_2=sk_dst_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + # breakpoint() + mu_dst, quat_dst=__DQB_warp2__( # [nu,3], [nu,4] + sk_w, # [N,K] + src_xyz_1, # [N,3] + sk_src_node_xyz_2, # [nu,n_neighbors,3] + sk_src_node_quat_2, # [nu,n_neighbors,4] + sk_dst_node_xyz_2, # [N,K,3] + sk_dst_node_quat_2, # [N,K,4] + dyn_o=None, + src_quat=src_wxyz_1, # [N,4] # wxyz + ) + src_xyz_1_new=mu_dst + src_xyzw_1_new=quat_dst[:,[1,2,3,0]] # [nu,4] + # breakpoint() + # for further optimization + blend_opt_dict = { + 'weight_param_key_optimizable': weight_param_key_optimizable, # (nu,n_neighbors) + 'src_xyz_1': src_xyz_1_new, # (nu,3) # CHECK + 'src_xyzw_1': src_xyzw_1_new, # (nu,4) # CHECK + 'per_nonkeyGaussian_dweight': per_nonkeyGaussian_dweight, # (nu,n_neighbors) + 'exp_sq_dists_key_nonkey_edges': exp_sq_dists_key_nonkey_edges, # (ne) + 'ts_most_confident': ts_most_confident, # (nu) + } + + # Step 2: calculate Transls_optimized and Oris_xyzw_optimized + Transls_optimized,Oris_xyzw_optimized=[],[] + ts_batches = torch.split(ts_all, batch_size) + for ts_batch in ts_batches: + nt=ts_batch.shape[0] + Mask_DQB=torch.ones_like(Mask[ts_batch],device=Mask.device) + # weight_param_key_optimizable=weight_param_key_optimizable.abs() + # weight_edges=cal_weight_edges2(weight_param_key_optimizable,exp_sq_dists_key_nonkey_edges,edges,nu) # [ne] # CHECK: 2000 + # weight_per_nonkeyGaussian=weight_edges.reshape(-1,n_neighbors) + sk_w=weight_per_nonkeyGaussian.expand(nt, -1, -1)[Mask_DQB] # (n_mask,n_neighbors) + + sk_src_node_xyz_1=Transls_key[ts_most_confident_repeat,edges[:,1]] # (ne,3) + sk_src_node_xyzw_1=Oris_xyzw_key_norm[ts_most_confident_repeat,edges[:,1]] # (ne,4) + sk_src_node_wxyz_1=sk_src_node_xyzw_1[:,[3,0,1,2]] + sk_src_node_xyz_2=sk_src_node_xyz_1.reshape(-1,n_neighbors,3) # (nu,n_neighbors,3) + sk_src_node_quat_2=sk_src_node_wxyz_1.reshape(-1,n_neighbors,4) # (nu,n_neighbors,4) + + sk_nt_src_node_xyz=sk_src_node_xyz_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_src_node_quat=sk_src_node_quat_2.expand(nt,-1,-1,-1)[Mask_DQB] # (n_mask,n_neighbors,4) + + # key points dst needed to be updated; src is the same + sk_nt_node_xyz_op_1=Transls_key_optimized[ts_batch][:,edges[:,1]] # (nt,ne,3) + sk_nt_node_xyz_op_2=sk_nt_node_xyz_op_1.reshape(nt,-1,n_neighbors,3) # (nt,nu,n_neighbors,3) + sk_nt_node_xyzw_op_1=Oris_xyzw_key_optimized_norm[ts_batch][:,edges[:,1]] # (nt,ne,4) + sk_nt_node_wxyz_op_1=sk_nt_node_xyzw_op_1[:,:,[3,0,1,2]] + sk_nt_node_quat_op_2=sk_nt_node_wxyz_op_1.reshape(nt,-1,n_neighbors,4) # (nt,nu,n_neighbors,4) + + sk_nt_dst_node_xyz_op=sk_nt_node_xyz_op_2[Mask_DQB] # (n_mask,n_neighbors,3) + sk_nt_dst_node_quat_op=sk_nt_node_quat_op_2[Mask_DQB] # (n_mask,n_neighbors,4) + + # nonkey points update when src_xyz_1 and src_xyzw_1 are trainable + src_xyz=src_xyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask,3) + src_wxyz_1=src_xyzw_1[:,[3,0,1,2]] + src_quat=src_wxyz_1.expand(nt,-1,-1)[Mask_DQB] # (n_mask<=nt*nu,4) + + # __LBS_warp2__ __DQB_warp2__ + mu_dst, quat_dst=__DQB_warp2__( # [n_mask,3], [n_mask,4] + sk_w, # [N,K] + src_xyz, # [N,3] + sk_nt_src_node_xyz, + sk_nt_src_node_quat, + sk_nt_dst_node_xyz_op, # [N,K,3] + sk_nt_dst_node_quat_op, # [N,K,4] + dyn_o=None, + src_quat=src_quat, # [N,4] # wxyz + ) + + quat_dst_xyzw=quat_dst[:,[1,2,3,0]] # [n_mask,4] + if not torch.allclose(quat_dst_xyzw.norm(dim=-1),torch.ones(quat_dst_xyzw.shape[0],device=quat_dst_xyzw.device)): + print("quat_dst_xyzw's norm != 1") + breakpoint() + Transls_optimized_batch=mu_dst.reshape(nt,nu,3) # [nt,nu,3] + Oris_xyzw_optimized_batch=quat_dst_xyzw.reshape(nt,nu,4) # [nt,nu,4] + Transls_optimized.append(Transls_optimized_batch) + Oris_xyzw_optimized.append(Oris_xyzw_optimized_batch) + Transls_optimized=torch.cat(Transls_optimized,dim=0) # [nt_all,nu,3] + Oris_xyzw_optimized=torch.cat(Oris_xyzw_optimized,dim=0) # [nt_all,nu,4] + + # ********************* new trying end ******************* + # breakpoint() + blend_dict={} + return Transls_optimized, Oris_xyzw_optimized, weight_param_key_optimizable,weight_distance_optimizable,blend_dict,blend_opt_dict + +def get_knn_edges_key(Transls,Mask,k): + # group by variance of relative motion + # Transls [nt,nk,3] + # std_threshold=0.01 + # max_dist_threshold=0.4 + + def nanstd(o, dim, keepdim=False): + result = torch.sqrt( + torch.nanmean( + torch.pow( torch.abs(o-torch.nanmean(o,dim=dim).unsqueeze(dim)),2), + dim=dim + ) + ) + if keepdim: + result = result.unsqueeze(dim) + return result + + nu=Transls.shape[1] + + # Dist=torch.cdist(Transls, Transls, p=2) + Dist = torch.norm(Transls[:, :, None, :] - Transls[:, None, :, :], dim=-1) # [nt, nu, nk] + Mask_dist=Mask[:, :, None] * Mask[:, None, :] # [nt, nu, nk] + Dist_masked = torch.where(Mask_dist, Dist, torch.tensor(float('nan'), device=Dist.device)) + # std_dist=Dist.std(dim=0) # [nu,nk] + std_dist=nanstd(Dist_masked,dim=0) # [nu,nk] + + mean_dist_masked=Dist_masked.nanmean(dim=0) # [nu,nk] # may include nan + # mean_dist=Dist.mean(dim=0) # [nu,nk] + Dist_masked2 = torch.where(Mask.unsqueeze(1).expand(-1, nu, -1), Dist, torch.tensor(float('nan'), device=Dist.device)) + mean_dist_masked2=Dist_masked2.nanmean(dim=0) # [nu,nk] # should not include nan # may still have nan + # mean_dist_masked2=Dist.mean(dim=0) # [nu,nk] # overwrite [TEST] + # breakpoint() + assert torch.any(mean_dist_masked2.isnan())==False + mean_dist_masked = torch.where(~torch.isnan(std_dist), mean_dist_masked, mean_dist_masked2) + # std_dist[~torch.isnan(std_dist)].mean(),mean_dist[~torch.isnan(mean_dist)].mean() + Dist_masked2 = torch.where(Mask_dist, Dist_masked, torch.tensor(float('-inf'), device=std_dist.device)) + # np.histogram(std_dist.numpy(),bins=50) + # np.histogram(std_dist[~torch.isnan(std_dist)].cpu().numpy(), bins=50) + # std_dist.min() + + max_dist = torch.max(Dist_masked2, dim=0).values # has -inf + max_dist2 = torch.where(~torch.isinf(max_dist), max_dist, torch.tensor(float('inf'), device=max_dist.device)) + # np.histogram(max_dist2.numpy()) + # np.histogram(max_dist[~torch.isinf(max_dist)].cpu().numpy(), bins=10) + + # (std_dist<0.01).sum(),(max_dist[~torch.isinf(max_dist)]<0.4).sum(),(max_dist2<0.4).sum(),torch.isnan(std_dist).sum() + + # try 1: + # connectivity_matrix=(std_dist40: + conditiona=Contribs2>0.5*np.max(Contribs2,axis=0) # 0.707 # TODO + # conditionb=Contribs2>contrib_threshold + # condition2=np.logical_or(conditiona,conditionb) + condition2=conditiona + else: + conditionb=Contribs2>contrib_threshold + condition2=conditionb + Mask2=torch.tensor(condition2,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable2=Oris_xyzw2.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edges + # 2.1 position edges + Dists_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True) + + # 2.2 orientation edges + Ori_distances_optimizable2=torch.rand(nu,nk,dtype=torch.float32,requires_grad=True)*2 + + # 3. edge weight + # 3.1 position stds + weight_distance_optimizable2=torch.zeros((nu,nk),dtype=torch.float32,requires_grad=True) + + # knn graph (max contribs version) + + # def o3d_knn_p(pts2,pts, num_knn): + # indices = [] + # sq_dists = [] + # pcd = o3d.geometry.PointCloud() + # pcd.points = o3d.utility.Vector3dVector(np.ascontiguousarray(pts, np.float64)) + # pcd_tree = o3d.geometry.KDTreeFlann(pcd) + # for p in pts2: + # [_, i, d] = pcd_tree.search_knn_vector_3d(p, num_knn) + # indices.append(i[0:]) + # sq_dists.append(d[0:]) + # return np.array(sq_dists), np.array(indices) + + + # i_s_knn_others=[] + # sq_dists_5=[] + # n_neighbors=5 # 5 # 10 can not work + # ts_most_confident=[] + # for i in range(nu): + # t_most_confident=np.argmax(Contribs2[:,i]) # np.argsort(Contribs2[:,i])[-1] + # eff_map=Contribs[t_most_confident]>contrib_threshold # key points map at t_most_confident + # pts=Transls[t_most_confident][eff_map] # confident key points at t_most_confident + # i_s=torch.arange(nk)[eff_map] # confident key points indices at t_most_confident + # pt2=Transls2[t_most_confident,i] # i-th other points at t_most_confident + # pts2=[pt2] # just one point + # sq_dists, indices = o3d_knn_p(pts2, pts, n_neighbors) + # sq_dists,indices= sq_dists[0],indices[0] + + # flag_use_key_point_key_neighbors=True + # if flag_use_key_point_key_neighbors: + # ## given a key point, get the knn n_neighbors key points + # indice_nearest_key=indices[0] # nearest key point index + # pt3=pts[indice_nearest_key] # nearest key point + # pts3=[pt3] # just one key point + # sq_dists, indices = o3d_knn_p(pts3, pts, n_neighbors) # inlcude itself + # sq_dists,indices= sq_dists[0],indices[0] + # ## + # # calcuate the distance between the i-th other point and the neighboring key points + # sq_dists=np.linalg.norm(pts[indices]-pt2,axis=-1) # [n_neighbors] + + + # i_s_knn=i_s[indices] # usage: ids[i_s_knn] + # i_s_knn_others.append(i_s_knn) + # sq_dists_5.append(sq_dists) + # ts_most_confident.append(t_most_confident) + # i_s_knn_others=torch.stack(i_s_knn_others) + # sq_dists_5=np.array(sq_dists_5) + # sq_dists_5=torch.tensor(sq_dists_5,dtype=torch.float32) + # ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + # edges_others=torch.concatenate([torch.arange(nu).unsqueeze(1).repeat(1,n_neighbors).reshape(-1)[:,None],i_s_knn_others.reshape(-1)[:,None]],dim=-1) + # # edges_others.shape # nu*5 + + # knn graph (minmax distance + contrib version) + n_neighbors=5 + edges_others,mean_dists=get_knn_edges_others(Transls,Mask,Transls2,Mask2,k=n_neighbors) + sq_dists_5=mean_dists[edges_others[:,0],edges_others[:,1]].reshape(-1,n_neighbors) # [nu,n_neighbors] + # edges_others,dists_edges=get_knn_edges_naive(n_neighbors,Transls,Transls2,flag_remove_self=False) # ablation of uknn + # sq_dists_5=dists_edges.reshape(-1,n_neighbors) # [nu,n_neighbors] + ts_most_confident=np.argmax(Contribs2[:,:],axis=0) # [nu] + ts_most_confident=torch.tensor(ts_most_confident,dtype=torch.long) # [nu] + + + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + torch.cuda.empty_cache() + print(f"{torch.cuda.memory_allocated()/1024/1024/1024}GB") + + # Method 3: + # use Dual Quat Blending (refer to MoSca) + # sq_dists_key_nonkey=-torch.ones([nu,nk]) + # sq_dists_key_nonkey[edges_others[:,0],edges_others[:,1]]=sq_dists_5.reshape(-1) + sq_dists_key_nonkey_edges=sq_dists_5.reshape(-1) # [ne] + weight_param_key_optimizable=1*torch.ones(nk,dtype=torch.float32,requires_grad=True) # 2000 [TODO] + per_nonkeyGaussian_dweight=torch.zeros(nu,n_neighbors,dtype=torch.float32,requires_grad=True) + + # batch training along t + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") + + Transls_optimized2,Oris_xyzw_optimized2, weight_param_key_optimized2,weight_distance_optimized2,blend_dict,blend_opt_dict=\ + optimize_others_with_orientation_edges4(\ + edges_others,\ + Transls2, Transls_optimizable2.clone(), Mask2, \ + Oris_xyzw2, Oris_xyzw_optimizable2.clone(), \ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + weight_param_key_optimizable,sq_dists_key_nonkey_edges,\ + ts_most_confident,per_nonkeyGaussian_dweight,\ + W2Cs,Depths_info_perG_nonkey, + device) + + + # keep + edges_others_optimized=edges_others + Dists_optimized2=Dists_optimizable2 + Ori_distances_optimized2=Ori_distances_optimizable2 + + return Transls_optimized2,Dists_optimized2,weight_distance_optimized2,Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others,weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict + + + + + + +def main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges,threshold_min_set): + import sys + import os + sys.path.append(os.path.dirname("../flow3d")) + + import torch + import numpy as np + from scipy.spatial.transform import Rotation + import matplotlib.pyplot as plt + import pickle + import random + import glob + + import pypose as pp + from torch import nn + + from itertools import chain + from scipy.spatial import KDTree,Delaunay,distance_matrix + import open3d as o3d + from collections import defaultdict + import roma + + # set args + + # work_dir="/data/dataset/used_ds/som/sriracha-tree/work-dir" # backpack,haru-sit, sriracha-tree + # work_dir="/data/dataset/used_ds/som/backpack/work-dir" # work-dir1 # initialization results (no SoM model) + # work_dir="/data/dataset/custom_ds_trained/rhino" + # work_dir="/data/dataset/custom_ds_graph_model/rhino" + work_dir=save_dir + + # file_effective="effective_dict.pkl" # backpack 854 + # file_effective="effective_dict_949.pkl" # haru-sit + # file_effective="effective_dict_1908.pkl" # sriracha-tree # 1908 # 1000 + # file_effective="effective_dict_484.pkl" # backpack # work-dir1 + # file_effective="effective_dict_712.pkl" # rhino + + # dict_dir=work_dir+"/out_dicts2" # has w2c and camera's K + dict_dir=work_dir+"/out_dicts3" # has w2c and camera's K + + flag_draw_contribs_sample=False + flag_visulize_sparisity=False + flag_visulize_key_point=False + flag_visulize_nonkey_point=False + + flag_auto_threshold=False + contrib_threshold=2 # usually 2 + threshold=2e-0 ## TODO # for non-key points # fixed threshold + # threshold=-torch.inf + + + + # [1] load data + nt=count_out_dict_files(dict_dir) + dicts=[] + for t in range(nt): + with open(f"{dict_dir}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # print(dicts[0].keys()) + + with open(f"{dict_dir}/{file_effective}", 'rb') as file: + effective_dict = pickle.load(file) + inds_effective = effective_dict['ids_eff2'] #ids_eff1 + # Contribs_up = effective_dict['Contribs_up'] + threshold_min = effective_dict['threshold_min'] + threshold_mean = effective_dict['threshold_mean'] + Contribs_all = effective_dict['Contribs'] + print(f"threshold_min:{threshold_min}, threshold_mean:{threshold_mean}") + + # threshold_min=np.max([threshold_min-1, 0.5]) # 0.5 + # threshold_min=np.max([threshold_min-3, 0.5]) # 0.5 # paper-windmill test + # threshold_min=0.1 # test: believe nearly all points # paper-windmill + # threshold_min=0.5 # test: camel,backpack + if threshold_min_set!=-1: + print(f"set threshold_min to {threshold_min_set}.") + threshold_min=threshold_min_set + else: + print(f"use threshold_min from effective_dict: {threshold_min}.") + + # breakpoint() + if flag_auto_threshold: + print(f"set auto threshold to {threshold_min}.") + contrib_threshold=threshold_min + threshold=threshold_min + # contrib_threshold=threshold_mean + # threshold=threshold_mean + + print(f"set threshold_min to {threshold_min}.") + + device='cuda' + ids=inds_effective + # ids=np.array(inds_effective)[np.arange(0,200,6)].tolist() # quick test + # ids=np.array(inds_effective)[np.arange(0,854,2)].tolist() + + i_target=0 + i_show=10 + Transls=np.array([dicts[t]["means"][ids].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz=np.array([dicts[t]["quats"][ids].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw=Quats_wxyz[...,[1,2,3,0]] # xyzw + # Contribs=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) #### + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids] for t in range(nt)]) #### + Contribs=Contribs_all[:,ids] + # Contribs=Contribs_up[:,ids] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort=np.array([get_order(Contribs[:,i]) for i in range(Contribs.shape[1])]).transpose(1,0) + # Vars=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids] for t in range(nt)]) + W2Cs=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + SE3s=np.concatenate([Transls,Quats_xyzw],axis=-1) # xyz xyzw (required by pypose) + SE3s=torch.tensor(SE3s,dtype=torch.float32) #,device='cuda') + Oris_xyzw=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type) + Oris=pp.LieTensor(SE3s[:,:,3:],ltype=pp.SO3_type).Log() + lie_tensor_SE3s=pp.LieTensor(SE3s, ltype=pp.SE3_type) + Depths_info_perG=torch.stack([dicts[t]['depths_info_perG'].cpu() for t in range(nt)]) + # depth_ratio=0.001 # 0.001 + Depths_info_perG=depth_ratio*torch.ones_like(Depths_info_perG) # TEST + print(f"TEST: set Depths_info_perG={depth_ratio}*torch.ones_like(Depths_info_perG).") # test + Depths_info_perG_key=Depths_info_perG[:,ids] + # len(dicts) + # vars=Vars[:,i_target] + nk=len(ids) # num of key points + # np.histogram(vars[vars<1e0]) + print("nk:",nk) + print("dicts[0]['means'].shape:", dicts[0]["means"].shape) + + # draw + if flag_draw_contribs_sample: + # i_show=20 + # means=Transls[:,i_show] + # vars0=Vars[:,i_show] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars0,axis=1)),cmap='viridis') + + # means=Transls[:,i_target] + # vars1=Vars[:,i_target] + # # sc=ax.scatter(means[:,0], means[:,1], means[:,2],s=1,alpha=0.6,c=np.log(np.sum(vars1,axis=1)),cmap='viridis') + + plt.figure(figsize=(10,4)) + # plt.subplot(1,3,1) + # plt.plot(vars0,c='r') + # plt.plot(vars1,c='b',alpha=0.6) + # plt.xlabel("time frame") + # plt.ylabel("uncertainty") + # plt.ylim([0,1]) + i_show=123 + plt.subplot(1,3,2) + plt.plot(Contribs[:,i_show],c='r') + # plt.plot(Contribs[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution") + plt.xlabel("time frame") + plt.subplot(1,3,3) + plt.plot(Contribs_sort[:,i_show],c='r') + # plt.plot(Contribs_sort[:,i_target],c='b',alpha=0.6) + plt.ylabel("contribution sort") + plt.xlabel("time frame") + plt.show() + + + # [2] Graph initialization for key Gaussians + # Initializations + # 0. naive ori distance + Oris_rel=calculate_edge_relative_orientation(Oris_xyzw) + Ori_distances_naive=get_navie_ori_distances(Oris_rel) # (nk, nk) + + # 1. vertices + # 1.1 positions + Transls=torch.tensor(Transls,dtype=torch.float32) + Transls_optimizable=Transls.clone().requires_grad_(True) + # Transls_optimizable=torch.tensor(Transls,dtype=torch.float64,requires_grad=True) + # Transls_optimizable=torch.nn.Parameter(Transls_optimizable) + # condition1a=Contribs>0.4*np.max(Contribs,axis=0) # 0.707 + # # condition1b=Contribs2>contrib_threshold + # # condition1=np.logical_or(condition1a,condition1b) + # condition1=condition1a + # Mask=torch.tensor(condition1,dtype=torch.bool) + Mask=torch.tensor(Contribs>contrib_threshold,dtype=torch.bool) + + # # 1.2 orientations + Oris_xyzw_optimizable=Oris_xyzw.clone().requires_grad_(True) # LieTensor SO3 + + # 2. edge distance + Dists = np.array([distance_matrix(Transls[t], Transls[t]) for t in range(nt)]) + Dists_optimizable=np.mean(Dists,axis=0) + Dists_optimizable=torch.tensor(Dists_optimizable,dtype=torch.float32,requires_grad=True) + + Ori_distances_optimizable=Ori_distances_naive.clone() + Ori_distances_optimizable=torch.tensor(Ori_distances_optimizable,dtype=torch.float32,requires_grad=True) + + # 3. edge weight + weight_distance_optimizable=1*torch.ones(Dists_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_distance_optimizable=weight_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # use the different weight + weight_ori_distance_optimizable=1*torch.ones(Ori_distances_optimizable.shape,dtype=torch.float32,requires_grad=True) + weight_ori_distance_optimizable=weight_ori_distance_optimizable.fill_diagonal_(1e-10)#(-100) # 1e-10 + + # 4. get local edges + n_neighbors=120 + if version_key_edges=="max_contrib": + edges_graph=get_knn_edges(n_neighbors,Transls,Contribs,Transls,Contribs,contrib_threshold,flag_remove_self=True) # edge version: max contrib version + elif version_key_edges=="minmax_distance_contrib": + # edges_graph,_=get_knn_edges_others(Transls,Mask,Transls,Mask,k=n_neighbors) # edge version: minmax distance + contrib mask version + edges_graph,_=get_knn_edges_key(Transls,Mask,k=n_neighbors) + elif version_key_edges=="max_native": + edges_graph,_=get_knn_edges_naive(n_neighbors,Transls,Transls,flag_remove_self=True) # edge version: max contrib version + # breakpoint() + else: + breakpoint() + raise ValueError("version_key_edges should be max_contrib or minmax_distance_contrib or max_native") + # edges_graph.shape # [12395, 2] + weight_distance_optimizable=torch.zeros(Dists_optimizable.shape,dtype=torch.float32) + weight_distance_optimizable[edges_graph[:,0],edges_graph[:,1]]=1 #1e-5 + weight_distance_optimizable.requires_grad=True + + # [3] optimization for key Guassians + print("===========================================") + print("Start Optimization for key Guassians") + print("-------------------------------------------") + + torch.cuda.empty_cache() + Transls_optimized,Dists_optimized,weight_distance_optimized,Oris_xyzw_optimized, Ori_distances_optimized=\ + optimize_overall_with_orientation_edges(\ + edges_graph,\ + Transls, Transls_optimizable.clone(), Mask, Dists_optimizable.clone(), weight_distance_optimizable.clone(),\ + Oris_xyzw, Oris_xyzw_optimizable.clone(), Ori_distances_optimizable.clone(), weight_ori_distance_optimizable, \ + None, None, None,None,\ + W2Cs,Depths_info_perG_key) + + print("-------------------------------------------") + print("Finish Optimization for key Guassians") + print("===========================================") + + if flag_visulize_sparisity: + # breakpoint() + weight_distance_optimized[2,:10],weight_distance_optimized[:10,2] # check symetry + # look at normed weight + torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None] + data = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=50) # check dist + # np.sort(data[4])[-100:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + # (data>1e-6).sum() + np.histogram((data>1e-6).sum(axis=-1)) + # ((data>1e-6).sum(axis=-1)==3).sum() + + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + # Example 2D array + print((data>0.02).sum()) + + # Create a meshgrid for the X and Y coordinates + x = np.arange(data.shape[1]) + y = np.arange(data.shape[0]) + x, y = np.meshgrid(x, y) + + # Create a figure and a 3D axis + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the surface + ax.plot_surface(x, y, np.flip(data,axis=1), cmap='viridis') + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_yticks([0,100,200,300,400]) + ax.set_yticklabels(nk-np.arange(0,500,100)) + ax.set_zlabel('Weight Values') + # ax.set_zlim(0,0.5) + ax.set_title('Edge Weights') + + plt.show() + + + if flag_visulize_key_point: + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=Transls.shape[0]-1 # 0 35 80 100 150 + # t=33 # test backpack + # t=0 # 306 block + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=2) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=2) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + # save + # get optimized graph from normed weights + norm_weight_distance_optimized = (torch.abs(weight_distance_optimized)/torch.norm(weight_distance_optimized,dim=-1,p=1)[:,None]).cpu().detach().numpy() + edges_graph_optimized=[] + for i in range(nk): + for j in range(nk): + if norm_weight_distance_optimized[i,j]>1e-6: + edges_graph_optimized.append([j,i]) + edges_graph_optimized=torch.tensor(edges_graph_optimized,dtype=torch.long) + + Ori_opt=Oris_xyzw_optimized + Nodes_opt=torch.concatenate([Transls_optimized.transpose(1,0),Ori_opt.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep=[[] for i in range(nk)] + # for e in edges_opt: + for e in edges_graph_optimized: + # Edges_keep[e[0]].append([e[1],e[0]]) + Edges_keep[e[1]].append([e[0],e[1]]) + + Edges_keep=[torch.tensor(Edges_keep[i],device=device) for i in range(nk)] + num_Edges_keep=[Edges_keep[i].shape for i in range(nk)] + # num_Edges_keep + + # check + torch.norm(Oris_xyzw_optimized[0],dim=-1)[:10] + + # for save + optimized_key_output={"Transls_optimized":Transls_optimized,\ + "Dists_optimized":Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized} + + # Save variables to a file + # torch.save({ + # 'dicts': dicts,\ + # 'ids': ids,\ + # 'Contribs': Contribs,\ + # 'Transls': Transls,\ + # 'Oris_xyzw': Oris_xyzw,\ + # 'Transls_optimizable': Transls_optimizable,\ + # 'Dists_optimizable': Dists_optimizable,\ + # 'weight_distance_optimizable':weight_distance_optimizable,\ + # 'Ori_distances_optimizable':Ori_distances_optimizable,\ + # 'Transls_optimized': Transls_optimized,\ + # 'Dists_optimized': Dists_optimized,\ + # "weight_distance_optimized":weight_distance_optimized,\ + # "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + # "Ori_distances_optimized":Ori_distances_optimized,\ + # "edges_graph":edges_graph,\ + # "edges_graph_optimized":edges_graph_optimized,\ + # "Nodes_opt":Nodes_opt,\ + # "Edges_keep":Edges_keep,\ + # "optimized_key_output":optimized_key_output,\ + # # "Transls_optimized1":Transls_optimized1,\ + # # "Dists_optimized1":Dists_optimized1,\ + # # "weight_distance_optimized1":weight_distance_optimized1,\ + # "device":device,\ + # "work_dir":work_dir}\ + + # , 'variables_8.pth') + + flag_read=False + if flag_read: + # Read variables from the file + checkpoint = torch.load('variables_8.pth') + dicts = checkpoint['dicts'] + ids = checkpoint['ids'] + Contribs = checkpoint['Contribs'] + Transls = checkpoint['Transls'] + Oris_xyzw = checkpoint['Oris_xyzw'] + Transls_optimized = checkpoint['Transls_optimized'] + Dists_optimized = checkpoint['Dists_optimized'] + weight_distance_optimized = checkpoint['weight_distance_optimized'] + Oris_xyzw_optimized = checkpoint['Oris_xyzw_optimized'] + Ori_distances_optimized = checkpoint['Ori_distances_optimized'] + edges_graph = checkpoint['edges_graph'] + edges_graph_optimized=checkpoint['edges_graph_optimized'] + Nodes_opt = checkpoint['Nodes_opt'] + Edges_keep = checkpoint['Edges_keep'] + optimized_key_output = checkpoint['optimized_key_output'] + # Transls_optimized1 = checkpoint['Transls_optimized1'] + # Dists_optimized1 = checkpoint['Dists_optimized1'] + # weight_distance_optimized1 = checkpoint['weight_distance_optimized1'] + device=checkpoint['device'] + work_dir=checkpoint['work_dir'] + nk=len(ids) + + + # [4] Graph initialization for non-key Gaussians + # 1. prepare data for non-key Gaussians + Ng=dicts[0]["means"].shape[0] # num of all Gaussians + nt=len(dicts) + # Ng=1000 + ids2=[i for i in range(Ng) if i not in ids] # ids of non-target nodes + # Contribs2_temp=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2_temp=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2_temp=Contribs_all[:,ids2] + Contribs2_temp_max=Contribs2_temp.max(axis=0) + if nt>350: + n_nonkey_max=36000 # 36000 # 150000 72000 too large memory for block teddy wheel + elif nt>40: + n_nonkey_max=72000 + else: + n_nonkey_max=150000 + if Contribs2_temp_max.shape[0]>n_nonkey_max: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[-n_nonkey_max] # the nth (12000th/24000th) largest contrib value + else: + contribs2_nth_largest = np.sort(Contribs2_temp_max)[0] + if nt>40: # iphone davis + contribs2_nth_largest=max(contribs2_nth_largest,0.01) # 0.1 + else: # nvidia + contribs2_nth_largest=max(contribs2_nth_largest,1e-4) + threshold=contribs2_nth_largest # rewrite threshold + ids2=np.array(ids2)[Contribs2_temp.max(axis=0)>threshold].tolist() # remove nodes with small contribs #15 + + + Transls2=np.array([dicts[t]["means"][ids2].cpu().numpy() for t in range(nt)]) #(t,|ids|,3) + Quats_wxyz2=np.array([dicts[t]["quats"][ids2].cpu().numpy() for t in range(nt)]) # wxyz (t,|ids|,4) + Quats_xyzw2=Quats_wxyz2[...,[1,2,3,0]] # xyzw + # Contribs2=np.array([dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # Contribs2=np.array([dicts[t]['contribs_in_mask'].cpu().numpy()[ids2] for t in range(nt)]) + Contribs2=Contribs_all[:,ids2] + # Contribs_sort=np.array([np.argsort(Contribs[t],axis=-1) for t in range(nt)]) + Contribs_sort2=np.array([get_order(Contribs2[:,i]) for i in range(Contribs2.shape[1])]).transpose(1,0) + # Vars2=np.array([1/dicts[t]['contribs'].reshape(-1,3).cpu().numpy()[:,0][ids2] for t in range(nt)]) + # W2Cs2=np.array([dicts[t]["w2c"].cpu().numpy() for t in range(nt)]) + # SE3s2=np.concatenate([Transls2,Quats_xyzw2],axis=-1) # xyz xyzw (required by pypose) + # SE3s2=torch.tensor(SE3s2,dtype=torch.float32) #,device='cuda') + Oris_xyzw2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type) + Oris2=pp.LieTensor(Quats_xyzw2,ltype=pp.SO3_type).Log() + Depths_info_perG_nonkey=Depths_info_perG[:,ids2] + nu=len(ids2) # num of non-target nodes + print(f'Ng={Ng},nk=len(ids)={len(ids)},nu=len(ids2)={len(ids2)},threshold={threshold}') + + # # test threshold + # condition1=Contribs2>0.707*np.max(Contribs2,axis=0) + # condition2=Contribs2>contrib_threshold + # np.histogram((np.logical_or(condition1,condition1)).sum(axis=0),bins=30) + + # 2. run optimizaiton for non-key Guassians + print("===========================================") + print("Start Optimization for nonkey Guassians") + print("-------------------------------------------") + + Transls_optimized2,Dists_optimized2,weight_distance_optimized2,\ + Oris_xyzw_optimized2, Ori_distances_optimized2,edges_others_optimized,edges_others, \ + weight_param_key_optimized2,ts_most_confident,blend_dict,blend_opt_dict \ + =run_batch(Transls2,Oris_xyzw2,Contribs2,\ + Transls,Oris_xyzw,Transls_optimized,Oris_xyzw_optimized,\ + Contribs,contrib_threshold,\ + W2Cs,Depths_info_perG_nonkey,\ + Mask) + + print("weight_param_key_optimized2 hist:",np.histogram(weight_param_key_optimized2.cpu().detach().numpy(),bins=10)) + + print("Finish Optimization for nonkey Guassians") + print("===========================================") + + + + if flag_visulize_nonkey_point: + # 3. check + # just for Method 1 + data = (torch.abs(weight_distance_optimized2)/torch.norm(weight_distance_optimized2,dim=-1,p=1)[:,None]).cpu().detach().numpy() # Replace with your actual data + # np.histogram(data[2],bins=5) # check dist + np.sort(data[4])[-5:] # check several largest weight + # np.argsort(data[2])[-100:] # check several largest weight id + # (np.diag(data)>1e-6).sum() # check diagonal + print((data>1e-6).sum(axis=-1)[:50]) # check sparsity # lower than 1e-7 or 1e-6 mean no meaning contribution + np.histogram((data>1e-6).sum(axis=-1)) # check sparsity + + import matplotlib.pyplot as plt + from mpl_toolkits.mplot3d import Axes3D + import numpy as np + + t=nt-1 # 0 35 80 100 150 + # t = 168 # backpack test + # t = 0 # 306 block # 33 + incorrect_vertex_ids=np.array([i for i in range(nk) if Contribs[t,i]contrib_threshold]) + + pts1=Transls_optimized.cpu().detach().numpy()[t] + pts2=Transls.cpu().numpy()[t] + pts=pts1 + pts_others=Transls_optimized2.cpu().detach().numpy()[t] + # pts_others=Transls2.cpu().detach().numpy()[t] + + # Generate some random 3D points for two arrays + x1 = pts[correct_vertex_ids][:,0] + y1 = pts[correct_vertex_ids][:,1] + z1 = pts[correct_vertex_ids][:,2] + + x2 = pts[incorrect_vertex_ids][:,0] + y2 = pts[incorrect_vertex_ids][:,1] + z2 = pts[incorrect_vertex_ids][:,2] + + x3 = pts_others[:,0] + y3 = pts_others[:,1] + z3 = pts_others[:,2] + + # Create a 3D plot + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + # Plot the first set of points in blue + ax.scatter(x1, y1, z1, c='b', label='correct_vertex_ids Points',s=4) + + # Plot the second set of points in red + ax.scatter(x2, y2, z2, c='r', label='incorrect_vertex_ids Points',s=4) + + ax.scatter(x3, y3, z3, c='k', label='other Points',s=1) + + # Add labels + ax.set_xlabel('X-axis') + ax.set_ylabel('Y-axis') + ax.set_zlabel('Z-axis') + # ax.set_box_aspect([1.0, 1.0, 1.0]) + ax.axis('equal') + # Add legend + ax.legend() + + # Show the plot + plt.show() + + Ori_opt2=torch.tensor(Quats_xyzw2,device=device) # tempory + # Ori_opt2=Oris_xyzw_optimized2.tensor() + Nodes_opt2=torch.concatenate([Transls_optimized2.transpose(1,0),Ori_opt2.transpose(1,0)],dim=-1) # (nk,nt,7) + Edges_keep2=[[] for i in range(nu)] + # for e in edges_opt: + for e in edges_others_optimized: + Edges_keep2[e[0]].append([e[1],e[0]+nk]) # [key id, other id + offset (nk)] + # Edges_keep2[e[1]].append([e[0],e[1]]) + + Edges_keep2=[torch.tensor(Edges_keep2[i],device=device) for i in range(nu)] + num_Edges_keep2=[Edges_keep2[i].shape for i in range(nu)] + + # combine key and others + Nodes_opt_combine=torch.cat([Nodes_opt,Nodes_opt2],dim=0) + Edges_keep_combine=Edges_keep+Edges_keep2 + ids_combine=ids+ids2 + + optimized_others_output={"Transls_optimized":Transls_optimized2,\ + "Dists_optimized":Dists_optimized2,\ + "weight_distance_optimized":weight_distance_optimized2,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized2, \ + "Ori_distances_optimized":Ori_distances_optimized2} + + + # [5] Save graph preprocessing data + quats_xyzw=Nodes_opt[:,:,3:] # xyzw + quats_wxyz=quats_xyzw[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt=[Nodes_opt[:,:,:3],quats_wxyz] # [[nk,nt,3], [nk,nt,4]] + + # combine + quats_xyzw_combine=Nodes_opt_combine[:,:,3:] # xyzw + quats_xyzw_combine=quats_xyzw_combine[..., [3, 0, 1, 2]] # wxyz + tracks_SE3_opt_combine=[Nodes_opt_combine[:,:,:3],quats_xyzw_combine] # [[nk,nt,3], [nk,nt,4]] + + # tracks_SE3_opt[1].shape,Nodes_opt.shape,tracks_SE3_opt_combine[1].shape,Nodes_opt_combine.shape + + nk=len(Edges_keep) + adjacent_matrix=torch.zeros((nk,nk)) + for i in range(nk): + if Edges_keep[i].shape[0]==0: + continue + indices=Edges_keep[i][:,0] + adjacent_matrix[i,indices]=1 + + adjacent_matrix = torch.logical_or(adjacent_matrix.t(), adjacent_matrix).type(torch.int32) + + Edges_keep_bi=[] + for i in range(nk): + indices=torch.where(adjacent_matrix[i,:]==1)[0] + edges_keep_bi=torch.cat((indices.unsqueeze(0),i*torch.ones_like(indices).unsqueeze(0))).t() + Edges_keep_bi.append(edges_keep_bi) + + Edges_keep_bi_combine=Edges_keep_bi+Edges_keep2 + + + # Save 1: + is_target=list(range(len(ids))) + is_target_combine=list(range(len(ids_combine))) + opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + "optimized_key_output":optimized_key_output,\ + "Nodes_opt_combine":Nodes_opt_combine,"tracks_SE3_opt_combine":tracks_SE3_opt_combine,"ids_combine":np.array(ids_combine),\ + "is_target_combine":is_target_combine,"Edges_keep_combine":Edges_keep_combine,"Edges_keep_bi_combine":Edges_keep_bi_combine,\ + "optimized_others_output":optimized_others_output} + # opt_dict={"Nodes_opt":Nodes_opt,"tracks_SE3_opt":tracks_SE3_opt,"ids":np.array(ids),\ + # "is_target":is_target,"Edges_keep":Edges_keep,"Edges_keep_bi":Edges_keep_bi,\ + # "optimized_key_output":optimized_key_output} + import pickle + dict_dir=f"{work_dir}/opt_dicts" + os.makedirs(dict_dir, exist_ok=True) + file_name=f"opt_v8_{len(ids)}_0.pkl" + # file_name=f"opt_427_var0.002_cut0.002_2.pkl" # 854 eff file + # file_name=f"opt_854_var0.002_cut0.002_nearby.pkl" # 854 eff file + # file_name=f"opt_2544_var0.002_cut0.002.pkl" # 2544 eff file + # file_name=f"opt_2038_var0.002_cut0.002_test.pkl" # 2544 eff file + # file_name="opt_dicts_test3.pkl" + + # # save + # file=open(f'{dict_dir}/{file_name}', 'wb') + # pickle.dump(opt_dict, file) + # file.close() + # print("file_path:",f'{dict_dir}/{file_name}') + + # Save 2: + # Save variables to a file + dict_dir_graph=f"{work_dir}/graph_model/" + os.makedirs(dict_dir_graph, exist_ok=True) + # file_name_graph=f"v8_712_parameters0.pth" + file_path_graph=dict_dir_graph+file_name_graph + + torch.save({ + 'dicts': dicts,\ + 'ids': ids,\ + 'Contribs': Contribs,\ + 'Transls': Transls,\ + 'Oris_xyzw': Oris_xyzw,\ + 'Transls_optimizable': Transls_optimizable,\ + 'Dists_optimizable': Dists_optimizable,\ + 'weight_distance_optimizable':weight_distance_optimizable,\ + 'Ori_distances_optimizable':Ori_distances_optimizable,\ + 'Transls_optimized': Transls_optimized,\ + 'Dists_optimized': Dists_optimized,\ + "weight_distance_optimized":weight_distance_optimized,\ + "Oris_xyzw_optimized":Oris_xyzw_optimized, \ + "Ori_distances_optimized":Ori_distances_optimized,\ + "edges_graph": edges_graph,\ + "edges_graph_optimized": edges_graph_optimized,\ + "Nodes_opt":Nodes_opt,\ + "Edges_keep":Edges_keep,\ + "optimized_key_output":optimized_key_output,\ + # "Transls_optimized1":Transls_optimized1,\ + # "Dists_optimized1":Dists_optimized1,\ + # "weight_distance_optimized1":weight_distance_optimized1,\ + "device":device,\ + "work_dir":work_dir,\ + "ids2":ids2,\ + "Contribs2": Contribs2,\ + "Transls_optimized2":Transls_optimized2,\ + "Dists_optimized2":Dists_optimized2,\ + "weight_distance_optimized2":weight_distance_optimized2,\ + "Oris_xyzw_optimized2":Oris_xyzw_optimized2,\ + "Ori_distances_optimized2":Ori_distances_optimized2,\ + "edges_others_optimized":edges_others_optimized,\ + "edges_others":edges_others,\ + "blend_opt_dict":blend_opt_dict,\ + "contrib_threshold":contrib_threshold,\ + }\ + , file_path_graph) + + print("file_path_graph:",file_path_graph) + + +def main2(save_dir,voxel_size=0.4,n_min_effective_frames=5,ratio_key=-1): + flag_remove_outlier=False + flag_show_sample=False + flag_show_threshold=False + flag_ablation_select_pts_wo_uncertainty=False + flag_ablation_random_select=False + # ratio_key=0.08 # ablation study of threshold and selected number of pts at each step. + + + # folder_dict="/data/dataset/used_ds/som/sriracha-tree/work-dir/out_dicts2" # haru-sit, backpack, sriracha-tree + # folder_dict="/data/dataset/custom_ds_graph_model/rhino/out_dicts3" # rhino + folder_dict=f"{save_dir}/out_dicts3" + + nt=count_out_dict_files(folder_dict) + dicts=[] + for t in range(nt): + with open(f"{folder_dict}/out_dict_{t}.pkl", 'rb') as file: + loaded_dict = pickle.load(file) + dicts.append(loaded_dict) + + # Contribs=np.array([dicts[t]['contribs_in_mask'].cpu().numpy() for t in range(nt)]) #### (old) + Contribs=np.array([dicts[t]['contribs_strict'].cpu().numpy() for t in range(nt)]) #### + + # Safety check for empty Contribs + if Contribs.size == 0: + breakpoint() + print("Warning: Contribs array is empty, using contribs_in_mask as fallback") + + if ratio_key>0: + max_num_key_nodes=int(Contribs.shape[1]*ratio_key) + num_stage_wise_select_key_nodes=int(max_num_key_nodes/1.5) + max_num_key_nodes = min(max_num_key_nodes, 1500) # avoid used memory is too large. + max_num_key_nodes = max(max_num_key_nodes, 5) # avoid no of key nodes too few. + else: + max_num_key_nodes=1500 # preivous setting + num_stage_wise_select_key_nodes=1000 # previous setting + + print("******ratio_key: ",ratio_key, "max_num_key_nodes: ",max_num_key_nodes," num_stage_wise_select_key_nodes: ",num_stage_wise_select_key_nodes) + + + c99=np.quantile(Contribs, 0.99) + c80=np.quantile(Contribs, 0.80) + contrib_ref=Contribs[(Contribsc80)].mean() + if contrib_ref<0.5: + prod_num=0.5/contrib_ref + Contribs=Contribs*prod_num + Contribs = np.clip(Contribs, a_min=None, a_max=10) + # breakpoint() + # dicts[0].keys(),nt,dicts[0]['means'].shape + + if flag_remove_outlier: + def get_outliers_by_density(means): + X=means[np.arange(0,means.shape[0],1),:] + + # Apply HDBSCAN clustering + # clusterer = hdbscan.HDBSCAN(min_cluster_size=10, min_samples=5, allow_single_cluster=True) + # labels = clusterer.fit_predict(X) + clusterer = DBSCAN(eps=0.02, min_samples=10) ############### hyperparameter!!! + labels=clusterer.fit(X).labels_ + outlier_map=labels==-1 + + # Unique clusters + # n_clusters = len(set(labels)) - (1 if -1 in labels else 0) # Ignore noise label (-1) + # print(f"Detected Clusters: {n_clusters}") + # assert n_clusters==1 + + return outlier_map + + outlier_map=get_outliers_by_density(dicts[t]['means'].cpu().numpy()) + outlier_map.sum() + + Contribs_up=[] + for t in range(nt): + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + if flag_remove_outlier: + outlier_map=get_outliers_by_density(means) + contribs[outlier_map]=1e-10 ## + Contribs_up.append(contribs) + Contribs_up=np.array(Contribs_up) + else: + Contribs_up=None + + print("num of gaussians: ",dicts[0]['means'].shape[0]) + # 1st Gaussian selection round + # select according to contribs + selected_ids_ts=[] + for t in range(nt): + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + else: + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) # ascending order w.r.t. ids + n=contribs.shape[0] + # candidate_ids_map=sorted_indices>=n-1000 + candidates_in_raw_ids=sorted_indices[-num_stage_wise_select_key_nodes:] #1000 + # candidates_in_raw_ids=[i for i in range(n) if candidate_ids_map[i]] + + candidate_means=means[candidates_in_raw_ids] + # candidate_contribs=contribs[candidate_ids_map] + + # voxel_size = 0.4 # 0.5 # 0.4 + voxel_grid = np.floor(candidate_means / voxel_size) + _, unique_indices = np.unique(voxel_grid, axis=0, return_index=True) + selected_candidates_in_raw_ids=[candidates_in_raw_ids[i] for i in unique_indices] + if flag_ablation_random_select: + n_random_select=len(selected_candidates_in_raw_ids) + selected_candidates_in_raw_ids=np.random.choice(selected_candidates_in_raw_ids, n_random_select, replace=False).tolist() + selected_ids_ts.append(selected_candidates_in_raw_ids) + # print(f"t={t}, selected_candidates_in_raw_ids.shape={len(selected_candidates_in_raw_ids)}") + + counts=np.zeros(n) + for t in range(nt): + selected_candidates_in_raw_ids=selected_ids_ts[t] + counts[selected_candidates_in_raw_ids]+=1 + + selected_ids_effective_1frame=[i for i in range(n) if counts[i]>0] + print("#pts left after 1st selection round: ", len(selected_ids_effective_1frame)) + # counts[selected_ids_effective_1frame] + + # avoid too many point in the 1st selection round + # select pts + def select_pts(n,selected_ids_effective_1frame): + arr1=np.ones(n) + arr2=np.zeros(len(selected_ids_effective_1frame)-n) + arr=np.concatenate((arr1,arr2),axis=0) + np.random.shuffle(arr) + selected_ids_effective_1frame=np.array(selected_ids_effective_1frame)[arr.astype(bool)].tolist() + return selected_ids_effective_1frame + # if len(selected_ids_effective_1frame)>2200: + # selected_ids_effective_1frame=select_pts(2200,selected_ids_effective_1frame) + # print("#pts left after 1st selection round (adjusted): ", len(selected_ids_effective_1frame)) + + # 2nd Gaussian selection round + ## 2nd selection round: whose contribs are effective at least nt/6 frames + # n_min_effective_frames=30 # 30 + # n_min_effective_frames=int(nt/6) + # n_min_effective_frames=min(5,int(nt/6)) # use 30 to avoid key pts not find in the long video. # TODO # 5 for haru-sit + print("n_min_effective_frames: ",n_min_effective_frames) + counts_effective_frames=np.zeros(len(selected_ids_effective_1frame)) + thresholds=[] + for t in range(nt): + if t%100==0: + print(t) + if not flag_remove_outlier: + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + else: + # means=dicts[t]['means'].cpu().numpy() + # outlier_map=get_outliers_by_density(means) + # contribs[outlier_map]=1e-10 ## + contribs=Contribs_up[t] + + sorted_indices=np.argsort(contribs) + # candidates_in_raw_ids=sorted_indices[-2000:] + threshold=contribs[sorted_indices[-num_stage_wise_select_key_nodes]] # the 2000th largest contrib # 1000 + threshold=max(threshold,0.5) # 0.5 + counts_effective_frames+=(contribs[selected_ids_effective_1frame]>=threshold).astype(int) + thresholds.append(threshold) + + threshold_min=np.min(np.array(thresholds)) + threshold_mean=np.mean(np.array(thresholds)) + print("min contrib threshold: ",threshold_min) + print("mean contrib threshold: ",threshold_mean) + print(np.histogram(counts_effective_frames)) + if flag_show_threshold: + import matplotlib.pyplot as plt + plt.figure() + plt.plot(thresholds) + plt.show() + selected_ids_eff=[selected_ids_effective_1frame[i] for i in range(len(selected_ids_effective_1frame)) if counts_effective_frames[i]>=n_min_effective_frames] + print("#pts left after 2nd selection round: ", len(selected_ids_eff)) + # counts_effective_frames[counts_effective_frames>30].shape + + # avoid too many point in the 2nd selection round + if len(selected_ids_eff)>max_num_key_nodes: # 1000 2500 # 1500 + selected_ids_eff=select_pts(max_num_key_nodes,selected_ids_eff) # 1000 2500 1500 + print("#pts left after 2nd selection round (adjusted): ", len(selected_ids_eff)) + + if flag_ablation_select_pts_wo_uncertainty: + selected_ids_eff=np.arange(Contribs.shape[1]) + selected_ids_eff=select_pts(1000,selected_ids_eff) # 1000 # 2500 1500 + print(f"flag_ablation_select_pts_wo_uncertainty={flag_ablation_select_pts_wo_uncertainty}") + print("random select key pts:", len(selected_ids_eff)) + + if flag_show_sample: + import matplotlib.pyplot as plt + # t=nt-1 + t=0 + # contribs2_contribs_id=dicts[t]['contribs'].reshape(-1,3).cpu().numpy() + # contribs=contribs2_contribs_id[:,0] # 0 or 1 or 2 + # contribs=dicts[t]['contribs_in_mask'].cpu().numpy() + contribs=Contribs[t] + means=dicts[t]['means'].cpu().numpy() + + fig = plt.figure() + ax = fig.add_subplot(111, projection='3d') + + x = means[selected_ids_eff][:,0] + y = means[selected_ids_eff][:,1] + z = means[selected_ids_eff][:,2] + # sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=np.log(contribs[selected_ids_eff]),cmap='viridis') + sc=ax.scatter(x, y, z,s=1,alpha=0.3,c=contribs[selected_ids_eff],cmap='viridis') + + cbar = plt.colorbar(sc) + cbar.set_label('Contribs') + ax.set_xlabel('X Label') + ax.set_ylabel('Y Label') + ax.set_zlabel('Z Label') + ax.axis('equal') + plt.show() + + # save + file_path=f"{folder_dict}/effective_dict_{len(selected_ids_eff)}.pkl" + effective_dict={"ids_eff2":selected_ids_eff,"ids_eff1":selected_ids_effective_1frame,"Contribs_up":Contribs_up, + "threshold_min":threshold_min,"threshold_mean":threshold_mean,"Contribs":Contribs} + file=open(file_path, 'wb') + pickle.dump(effective_dict, file) + file.close() + print(file_path) + file_effective=f"effective_dict_{len(selected_ids_eff)}.pkl" + return file_effective + +def compute_knn_indices(points, k, chunk_size=1024): + """ + Memory-efficient KNN search: returns indices of k nearest neighbors for each point. + points: [N, 3] + returns: [N, k] LongTensor + """ + N = points.shape[0] + knn_indices = [] + + for i in range(0, N, chunk_size): + end = min(i + chunk_size, N) + chunk = points[i:end] # [chunk, 3] + dists = torch.cdist(chunk, points) # [chunk, N] + topk = dists.topk(k + 1, largest=False).indices[:, 1:] # exclude self + knn_indices.append(topk) + + return torch.cat(knn_indices, dim=0) # [N, k] + +def gaussian_kde_density_knn(points, inv_covariances, k=1000, eps=1e-6): + """ + Vectorized version using k nearest neighbors for KDE density. + points: [N, 3] + covariances: [N, 3, 3] + returns: [N] density using k-NN Mahalanobis distances + """ + N, D = points.shape + device = points.device + + # Invert covariances: [N, 3, 3] + # cov_inv = torch.inverse(covariances + eps * torch.eye(D, device=device)) + cov_inv = inv_covariances + + # # Get k-NN indices using Euclidean distance + # dists = torch.cdist(points, points) # [N, N] + # knn_indices = dists.topk(k + 1, largest=False).indices[:, 1:] # [N, k] (exclude self) + + # Memory-efficient KNN + chunk_size= 10000 + knn_indices = compute_knn_indices(points, k, chunk_size=chunk_size) # [N, k] + + # Gather neighbor points: [N, k, 3] + knn_points = points[knn_indices] # batched gather + + # Center point per row: [N, 1, 3] -> broadcast to [N, k, 3] + diffs = knn_points - points[:, None, :] # [N, k, 3] + + # Apply Mahalanobis: dᵢⱼ = (xᵢ - xⱼ)^T Σᵢ⁻¹ (xᵢ - xⱼ) + # cov_inv: [N, 3, 3], diffs: [N, k, 3] + # compute: diffᵢⱼ^T @ cov_invᵢ @ diffᵢⱼ -> result [N, k] + + diffs_unsq = diffs.unsqueeze(-1) # [N, k, 3, 1] + cov_inv_exp = cov_inv[:, None, :, :] # [N, 1, 3, 3] + m = torch.matmul(cov_inv_exp, diffs_unsq) # [N, k, 3, 1] + mahalanobis_sq = torch.matmul(diffs_unsq.transpose(-1, -2), m).squeeze(-1).squeeze(-1) # [N, k] + + # Gaussian kernel weights + weights = torch.exp(-0.5 * mahalanobis_sq) # [N, k] + + density = weights.sum(dim=1) # [N] + return density + + +def plot_density_histogram_rgb(density, bins=20, height=256): + """ + density: [N] tensor + height: desired output image height in pixels + Returns: [height, width, 3] RGB uint8 numpy image of the histogram with log-scale x-axis and labeled axes + """ + density_np = density.cpu().numpy() + density_np = density_np[density_np > 0] # filter out non-positive for log scale + + dpi = 100 # Dots per inch + fig_height_inches = height / dpi + aspect_ratio = 1.5 # width / height + fig_width_inches = fig_height_inches * aspect_ratio + + # Create figure with desired size + fig, ax = plt.subplots(figsize=(fig_width_inches, fig_height_inches), dpi=dpi) + ax.hist(density_np, bins=bins, color='skyblue', edgecolor='black') + ax.set_xscale("log") + ax.set_title("Density Histogram (Log X-axis)") + ax.set_xlabel("Density (log scale)") + ax.set_ylabel("Count") + + fig.tight_layout() + fig.canvas.draw() + + width, height = fig.canvas.get_width_height() + image = np.frombuffer(fig.canvas.tostring_rgb(), dtype=np.uint8).reshape(height, width, 3) + + plt.close(fig) + return image.astype(np.uint8) + +def load_gt_masks(ws, gt_mask_dirname="mask"): + import imageio.v2 as imageio + """Load ground truth masks from directory""" + gt_mask_dir = osp.join(ws, gt_mask_dirname) + if not osp.exists(gt_mask_dir): + logging.warning(f"GT mask directory {gt_mask_dir} not found") + breakpoint() + return None + + gt_mask_fns = [ + f for f in os.listdir(gt_mask_dir) + if f.endswith(".png") or f.endswith(".jpg") + ] + gt_mask_fns.sort() + + gt_masks = [] + for gt_mask_fn in gt_mask_fns: + gt_mask_path = osp.join(gt_mask_dir, gt_mask_fn) + gt_mask = imageio.imread(gt_mask_path) + if gt_mask.ndim == 3: + gt_mask = gt_mask[..., 0] # Take first channel if RGB + gt_mask = (gt_mask > 128).astype(np.float32) # Convert to binary + gt_masks.append(gt_mask) + + gt_masks = torch.Tensor(np.stack(gt_masks)).float() # T,H,W + # self.register_gradfree_buffer("gt_masks", gt_masks) + # self.has_gt_masks = True + logging.info(f"Loaded GT masks from {gt_mask_dir}: {gt_masks.shape}, foreground ratio: {gt_masks.mean():.3f}") + return gt_masks, gt_mask_fns + +# @torch.inference_mode() +def main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model=None): + import torch + import argparse + import os + import torch + import time + from leo_renderer import Renderer + from lib_render.render_helper import render, render_gsplat + import numpy as np + from tqdm import tqdm + import imageio.v3 as iio + + from datetime import datetime + from loguru import logger as guru + import torch.nn.functional as F + import cv2 + import matplotlib.pyplot as plt + import lpips + + from lib_ugraph.ugraph_utils import compute_ssim_map, compute_lpips_map + + # parser = argparse.ArgumentParser() + # parser.add_argument("--cfg_fn", type=str) + # parser.add_argument("--port", type=int, default=8899) + # parser.add_argument("--work_dir", type=str, required=True) + # args = parser.parse_args() + + + + if dataset_name=='iphone': + cfg_fn='/data/repo/MoSca/profile/iphone/iphone_fit.yaml' + elif dataset_name=='davis' or dataset_name=='davis_mask': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='nvidia': + cfg_fn='/data/repo/MoSca/profile/nvidia/nvidia_fit.yaml' + elif dataset_name=='diffusion4d': + cfg_fn='/data/repo/MoSca/profile/demo/demo_fit.yaml' + elif dataset_name=='objaverse': + cfg_fn='/data/repo/MoSca/profile/objaverse/objaverse_fit.yaml' + else: + cfg_fn='' + breakpoint() + port=9000 + + device = torch.device("cuda") # if torch.cuda.is_available() else "cpu") + renderer = Renderer(cfg_fn, device, work_dir, port=port, bg_flag=True, fg_flag=True,load_s2d=True, ws=ws, use_ugraph=False, relative_dir_saved_mosca_model=relative_dir_saved_mosca_model) + + nt=renderer.cams.T + print("nt:",nt) + + W, H = renderer.cams.default_W, renderer.cams.default_H + assert W == renderer.s2d.W and H == renderer.s2d.H + K = renderer.cams.K(H=H,W=W) + # focal = 0.5 * H / np.tan(0.5 * camera_state.fov).item() + # K = torch.tensor( + # [[focal, 0.0, W / 2.0], [0.0, focal, H / 2.0], [0.0, 0.0, 1.0]], + # device=device, + # ) + # w2c = torch.linalg.inv( + # torch.from_numpy(camera_state.c2w.astype(np.float32)).to(self.device) + # ) + ts=torch.arange(nt,device=device) + w2cs=torch.stack([renderer.cams.T_cw(t) for t in ts],dim=0) + # with torch.inference_mode(): + # for w2c,t in tqdm(zip(w2cs,ts)): + # # w2c = renderer.cams.T_cw(t) + + # gs5 = [] + # if renderer.bg_flag: + # gs5.append(renderer.s_model()) + # if renderer.fg_flag: + # gs5.append(renderer.d_model(t)) + # render_dict, _ = render_gsplat( + # gs5, + # H, + # W, + # K=K, + # T_cw=w2c, + # bg_color=[1.0, 1.0, 1.0], + # ) + + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # # render_dict0 = render( + # # gs5, + # # H, + # # W, + # # K=K, + # # T_cw=w2c, + # # bg_color=[1.0, 1.0, 1.0], + # # ) + + # # img0 = torch.clamp(render_dict0["rgb"].permute(1, 2, 0), 0.0, 1.0) + # # img0 = (img0.cpu().numpy() * 255.0).astype(np.uint8) + # # combined = np.hstack([img0,img]) + # # iio.imwrite(f"./test/img_{t.item()}.png", combined) # check if they look the same. + # # compare_images(img, img0) + + if save_dir=='': + save_dir = work_dir + + video_dir = f"{save_dir}/videos/{datetime.now().strftime('%Y-%m-%d-%H%M%S')}" + print("*"*20) + print("*"*20) + print(f"Saving video to {video_dir}") + print("*"*20) + print("*"*20) + os.makedirs(video_dir, exist_ok=True) + + torch.backends.cuda.preferred_linalg_library("magma") + + # if dataset_name=="davis": + # prior_depths=torch.stack(train_dataset.depths).to(device) + # elif dataset_name=="iphone": + # prior_depths=train_dataset.depths.to(device) # [nt,H,W] + # else: + # # breakpoint() + # print("No prior depth!!!") + + if dataset_name=="davis" or dataset_name=="iphone" or dataset_name=="diffusion4d" or dataset_name=="objaverse" or dataset_name=="davis_mask": + thr_density_strict=0.1 #5 + thr_density_loose=0.1 #1 + thr_rgb_diff=0.3 + thr_ssim=0.01 + elif dataset_name=="nvidia": + thr_density_strict=0.0 + thr_density_loose=0.0 + thr_rgb_diff=0.6#0.6 + thr_ssim=0.001 # 0.001 + else: + breakpoint() + + try: + # gt_mask_dir = osp.join(ws, "mask") + # renderer.s2d.load_gt_masks(gt_mask_dir) + gt_masks=renderer.s2d.gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + except: + breakpoint() + # gt_masks=None + gt_masks, gt_mask_fns=load_gt_masks(ws, "mask") + gt_masks=gt_masks.to(device) + gt_masks=gt_masks.to(dtype=torch.bool) + # breakpoint() + prior_depths=renderer.s2d.dep.to(device) # [nt,H,W] + # breakpoint() + prior_depths_grad, _, _=compute_depth_gradient(prior_depths) # [nt,H,W] + # depths_info = 0.05*torch.exp(-30*prior_depths_grad) # [nt,H,W] # TODO: hyperparameter + # depths_info = depths_info.clamp_min(1e-4) # set minimum value to 1e-4 + depths_info = 1e-3 * torch.ones_like(prior_depths_grad,device=device) # 1e-3 1e-5 + # depths_info = 1e-3 * torch.ones((nt,H,W),device=device) # 1e-3 1e-5 + # depths_info[prior_depths_grad <= 0.03] = 0.1 # [nt,H,W] # TODO: hyperparameter # 0.05 # TODO!!! + print("np.histogram:",np.histogram(prior_depths_grad[gt_masks].flatten().cpu().numpy(), bins=10)) + # np.histogram(depths_info.flatten().cpu().numpy(), bins=10) + depth_min = prior_depths.min().item() + def safe_quantile(tensor: torch.Tensor, q: float, max_samples: int = 100000): + """Compute quantile safely by subsampling if needed.""" + if tensor.numel() > max_samples: + indices = torch.randperm(tensor.numel(), device=tensor.device)[:max_samples] + tensor = tensor.view(-1)[indices] + return tensor.quantile(q) + depth_max = safe_quantile(prior_depths, 0.99).item() + prior_depths_grad_threshold = safe_quantile(prior_depths_grad[gt_masks], 0.90).item() + print(f"set prior_depths_grad_threshold to {prior_depths_grad_threshold}.") + # lpips_model = lpips.LPIPS(net="alex", spatial=True).to(device) + # breakpoint() + print("fg Gaussian num:", renderer.d_model.N) + print("bg Gaussian num:", renderer.s_model.N) + print("generate out_dicts_t.pkl:") + for i, (w2c, t) in enumerate(zip(tqdm(w2cs), ts)): + assert i == t.item() + # if i<98: + # continue # test + with torch.inference_mode(): + # out_dict=renderer.model.render(int(t.item()), w2c[None], K[None], img_wh) # my code + gs5 = [] + if renderer.bg_flag: + gs5.append(renderer.s_model()) + if renderer.fg_flag: + d_list_t=renderer.d_model(t) + gs5.append(d_list_t) + n_fg = renderer.d_model(t)[0].shape[0] + render_dict, out_dict = render_gsplat( + gs5, + H, + W, + K=K, + T_cw=w2c, + bg_color=[1.0, 1.0, 1.0], + n_fg=n_fg, + ) + n_all=len(gs5) + # img = torch.clamp(render_dict["rgb"].permute(1, 2, 0), 0.0, 1.0) + # img = (img.cpu().numpy() * 255.0).astype(np.uint8) + + # density filter begin + mu, frame, scale, o, sph=d_list_t + # S = torch.zeros_like(frame) + # S[:, 0, 0] = scale[:, 0] + # S[:, 1, 1] = scale[:, 1] + # S[:, 2, 2] = scale[:, 2] + # actual_covariance = frame @ (S**2) @ frame.permute(0, 2, 1) + inv_actual_covariance = frame @ torch.diag_embed(1.0 / (scale ** 2)) @ frame.transpose(1, 2) + + # test = torch.distributions.MultivariateNormal(mu, actual_covariance).log_prob(mu) + density=gaussian_kde_density_knn(mu,inv_actual_covariance, k=100) # TODO: check scale factor + # breakpoint() + large_density_Gaussian_strict=density>thr_density_strict # TODO: threshold # 5 + large_density_Gaussian_loose=density>thr_density_loose # TODO: threshold # 1 + # density filter end + + out_dict['w2c']=w2c.detach() + out_dict['K']=K + # gt_mask = renderer.s2d.dep_mask[i].to(device) # (H, W) # test # not the true mask! + gt_mask = gt_masks[i] + # breakpoint() + # out_dict['gt_mask']=gt_mask + projected = project_points(out_dict['means'], K, w2c) # [N,2] + in_mask_Gaussians=valid_and_visible(projected, gt_mask) # [N] bool + out_dict['in_mask_Gaussians']=in_mask_Gaussians + contribs=out_dict['contribs'].reshape(-1,3)[:,0].clone() + + img = out_dict["img"][0] # [H,W,3] + gt_img = renderer.s2d.rgb[t].to(device) # [H,W,3] + diff_img = F.l1_loss(img, gt_img, reduction="none").mean(dim=-1) # [H,W] + mask_diff_img = diff_img < thr_rgb_diff # [H,W] # TODO: threshold #0.1 + small_img_diff_Gaussians=valid_and_visible(projected, mask_diff_img) # [N] bool + + # ssim map begin + ssim_map, ssim_loss=compute_ssim_map(img.cpu().numpy(),gt_img.cpu().numpy()) # [H,W,3] + mask_ssim_img = torch.tensor(np.array(ssim_map > thr_ssim),device=device).all(dim=-1) # [H,W] # TODO + large_ssim_Gaussians=valid_and_visible(projected, mask_ssim_img) # [N] bool + + # lpips_map, lpips_loss=compute_lpips_map(lpips_model, img.cpu().numpy(),gt_img.cpu().numpy(),device=device) # [H,W,1] + # mask_lpips_img = torch.tensor((lpips_map < 0.5)[:,:,0],device=device) # [H,W] # TODO + # small_lpips_Gaussians=valid_and_visible(projected, mask_lpips_img) # [N] bool + # breakpoint() + # ssim map end + # depth gradient mask + mask_depth_grad_img = prior_depths_grad[i] < prior_depths_grad_threshold # [H,W] + small_depth_grad_Gaussians=valid_and_visible(projected, mask_depth_grad_img) # [N] bool + + # + valid_gaussians_strict = small_depth_grad_Gaussians & large_ssim_Gaussians & small_img_diff_Gaussians & in_mask_Gaussians # & large_density_Gaussian_strict #& small_lpips_Gaussians # & in_mask_Gaussians # + valid_gaussians_loose = large_ssim_Gaussians & small_img_diff_Gaussians # & large_density_Gaussian_loose #& small_lpips_Gaussians # + # valid_gaussians=torch.ones(n_all, dtype=torch.bool, device=device) # TODO: threshold + # contribs_in_mask[~small_img_diff_Gaussians]=1e-8 # Gaussians on incorrect pixel are set to small contribs + contribs_in_mask=contribs.clone() + contribs_in_mask[~in_mask_Gaussians]=1e-8 # out-mask Gaussians are set to small contribs + contribs_strict=contribs.clone() + contribs_loose=contribs.clone() + contribs_strict[~valid_gaussians_strict]=1e-8 + contribs_loose[~valid_gaussians_loose]=1e-8 + + out_dict['contribs_in_mask']=contribs_in_mask + out_dict['contribs_strict']=contribs_strict + out_dict['contribs_loose']=contribs_loose + + # depth uncertainty + depths_info_perG=associate_pts_value(projected, depths_info[i]) # per Gaussian [N] + out_dict['depths_info_perG']=depths_info_perG + # out_dict['prior_depths_grad']=prior_depths_grad + + if True: + import pickle + dict_dir=f"{save_dir}/out_dicts3" + os.makedirs(dict_dir, exist_ok=True) + file=open(f'{dict_dir}/out_dict_{t}.pkl', 'wb') + pickle.dump(out_dict, file) + file.close() + + if True: + def to_rgb(img: np.ndarray) -> np.ndarray: + # [H, W] or [H, W, 1] -> [H, W, 3] + if img.ndim == 2: + return cv2.cvtColor(img, cv2.COLOR_GRAY2BGR) + elif img.ndim == 3 and img.shape[2] == 1: + return cv2.cvtColor(img[:, :, 0], cv2.COLOR_GRAY2BGR) + return img # Already RGB or unsupported format + + img_np = (img.cpu().numpy() * 255.0).astype(np.uint8) + gt_img_np = (gt_img.cpu().numpy() * 255.0).astype(np.uint8) + ssim_map_np = (np.array(ssim_map) * 255.0).clip(0, 255).astype(np.uint8) + # lpips_map_np = to_rgb((np.array(lpips_map) * 255.0).clip(0, 255).astype(np.uint8)) + + img_with_points = draw_projected_points_masked(img_np, projected, valid_gaussians_strict) # valid_gaussians, in_mask_Gaussians + img_with_points_loose = draw_projected_points_masked(img_np, projected, valid_gaussians_loose) + masked_img = overlay_mask_yellow(img_np, gt_mask.cpu().numpy()) + masked_gt_img = overlay_mask_yellow(gt_img_np, gt_mask.cpu().numpy()) + masked_diff_gt_img = overlay_mask_yellow(gt_img_np, mask_diff_img.cpu().numpy(), color= (255,0,0)) + + diff_img_np = ((diff_img*5).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # *10 to make it clear + diff_img_np = to_rgb(diff_img_np) + # depth_info_np= ((100*depths_info[t]).clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) # TODO: auto normalize for show + # depth_info_np = to_rgb(depth_info_np) + img_with_points_depth=draw_projected_points_colored(img_np, projected, depths_info_perG) + depth_img=apply_depth_colormap( + prior_depths[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = (depth_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + # depth_img_np = to_rgb(depth_img_np) + prior_depths_grad_img=apply_depth_colormap( + prior_depths_grad[i].unsqueeze(-1), near_plane=depth_min, far_plane=depth_max + ) + prior_depths_grad_img_np = (prior_depths_grad_img.clamp(0, 1).cpu().numpy() * 255.0).astype(np.uint8) + + # hist_img = plot_density_histogram_rgb(density,height=img_np.shape[0]) + + # combined = stack_images_side_by_side(img_with_points, masked_img) + # lpips_map_np, masked_img, masked_gt_img, hist_img , depth_info_np, depth_img_np, + combined = np.hstack([img_with_points, img_with_points_loose, depth_img_np, prior_depths_grad_img_np, ssim_map_np, masked_img, masked_gt_img, masked_diff_gt_img, diff_img_np, img_with_points_depth]) + iio.imwrite(f"{video_dir}/img_{i}.png", combined) + # breakpoint() + # del lpips_model.net + # lpips_model=None + # Delete large objects manually + del renderer + del render_dict + del inv_actual_covariance + # del lpips_model + del prior_depths + del prior_depths_grad + out_dict = {k: v.detach().cpu() if isinstance(v, torch.Tensor) else v for k, v in out_dict.items()} + del out_dict + del img, gt_img, diff_img, mask_diff_img + del projected, depths_info_perG + del contribs_strict, contribs_loose + del ssim_map, ssim_loss#, lpips_map, lpips_loss + + + # Optionally clear lists + gs5.clear() + + # Force garbage collection + import gc + gc.collect() + + # Clear PyTorch CUDA cache + torch.cuda.empty_cache() + # breakpoint() + +def compare_images(img, img0, tolerance=1): + if img.shape != img0.shape: + print(f"Shape mismatch: {img.shape} vs {img0.shape}") + return False + + diff = np.abs(img.astype(np.int32) - img0.astype(np.int32)) + max_diff = diff.max() + + if max_diff <= tolerance: + print(f"Images are close. Max diff: {max_diff}") + return True + else: + print(f"Images differ. Max diff: {max_diff}") + return False + +# def test(): +# # iphone +# # seq_name="block" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five, spin +# seq_name="backpack_" #"backpack_" +# # work_dir=f"/data/dataset/iphone_mosca_trained/paper-windmill/logs/iphone_fit_native_add3_20250226_115418" +# work_dir=f"/data/dataset/iphone_mosca_trained/backpack_/logs/iphone_fit_native_add3" #_20250223_123930" +# save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" +# dataset_name="iphone" + +# file_name_graph=f"v12_parameters8mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) +# main3(save_dir,file_effective,file_name_graph) + +# def test3(): +# # davis +# seq_name="camel" # rhino, camel, car-turn +# work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" +# save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" +# ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" +# # root_dir=f"/data/dataset/custom_mocsa_initialized" +# dataset_name="davis" + +# file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning +# # main1(dataset_name,work_dir,save_dir,ws) +# file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 +# main3(save_dir,file_effective,file_name_graph) + + +def test2(): + # iphone + # seq_names=['paper-windmill', 'block', 'teddy', 'apple', 'haru-sit']#, 'sriracha-tree', 'mochi-high-five', ] #,'sriracha-tree','block','teddy'] + # seq_name="mochi-high-five" # block, haru-sit, backpack, sriracha-tree, apple, paper-windmill, mochi-high-five + # 'backpack_', + # seq_names=['backpack_', 'apple', 'block', 'handwavy_', 'mochi-high-five_', 'pillow_', 'spin', 'teddy', 'creeper_', 'haru-sit_', 'paper-windmill', 'space-out', 'sriracha-tree_', 'wheel'] + # seq_names=['apple', 'block', 'spin', 'teddy', 'space-out', 'wheel'] + seq_names=['apple'] + + depth_ratio=0.1 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + # auto find the folder with timestamp + # base_dir = f"/data/dataset/iphone_mosca_trained/{seq_name}/logs" + # pattern = os.path.join(base_dir, "iphone_fit_native_add3*") + # matches = glob.glob(pattern) + # if matches: + # work_dir = matches[0] # or choose with more logic + # print("Found:", work_dir) + # else: + # print("No matching folders found.") + # print('='*20) + # continue + + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/logs/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + dataset_name="iphone" + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + # main1(dataset_name,work_dir,save_dir,ws) + torch.cuda.empty_cache() + file_effective=main2(save_dir,voxel_size=0.2,n_min_effective_frames=5) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="max_contrib",threshold_min=0.5) #"max_contrib" minmax_distance_contrib + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + + + +def test4(): + # davis # 'car-turn', 'camel', + # seq_names=['car-turn', 'bear', 'breakdance-flare', 'dance-twirl', 'drift-straight', 'helicopter', 'libby', 'mbike-santa', 'rhino', 'schoolgirls'] # + seq_names=['camel'] + # seq_names=['car-turn', 'bear', 'drift-straight', 'helicopter', 'mbike-santa', 'rhino'] # + depth_ratio=0.001 # 0.001 + for seq_name in seq_names: + print("=======================================================") + print("seq_name:", seq_name) + # try: + if True: + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/logs/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/" + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + dataset_name="davis" + + file_name_graph=f"v12_parameters8mee.pth" # o-opacity s-scale m-mask p-pruning + # main1(dataset_name,work_dir,save_dir,ws) + file_effective=main2(save_dir,voxel_size=0.05,n_min_effective_frames=5) # 0.2 + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges="minmax_distance_contrib",threshold_min=0.5) #"minmax_distance_contrib") + # except: + # print("-------------------------------------------------------") + # print(f"ERROR in {seq_name}") + print("=======================================================") + +def backup_code(save_dir): + from datetime import datetime + backup_dir = osp.join(save_dir, "src_backup") + os.makedirs(backup_dir, exist_ok=True) + for path in [ + "run_preprocessing.sh", + "graph_model_preprocessing5.py", + ]: + os.system(f"cp -r {path} {backup_dir}") + # reduce the backup size + # shutil.rmtree(osp.join(backup_dir, "lib_prior", "seg")) + # for root, dirs, files in os.walk(backup_dir): + # for file in files: + # if file.endswith(".pth") or file.endswith(".ckpt"): + # if osp.isfile(osp.join(root, file)): + # os.remove(osp.join(root, file)) + # else: + # shutil.rmtree(osp.join(root, file)) + # backu the commandline args + with open(osp.join(save_dir, f"commandline_args{datetime.now().strftime('%Y-%m-%d-%H%M%S')}.txt"), "w") as f: + f.write(" ".join(sys.argv)) + +def run_from_outside(): + parser = argparse.ArgumentParser() + parser.add_argument("--seq_name", type=str, required=True) + parser.add_argument("--func_name", type=str, required=True) + parser.add_argument("--dataset_name", type=str, required=True) + parser.add_argument("--depth_ratio", type=float, help="default=0.1 (iphone) or 0.001 (backpack or davis)", required=True) # default=0.1, + parser.add_argument("--threshold_min_set", type=float, help="default=0.5 or 0.1 (paper-windmill)", required=True) # default=0.5, + parser.add_argument("--version_key_edges", type=str, help="version_key_edges: max_contrib or minmax_distance_contrib", required=True) # default="max_contrib" + parser.add_argument("--extra_save_str", type=str, help="extra_save_str", required=False, default="") # _ab_uknn ## ablation study + parser.add_argument("--ratio_key", type=float, help="ratio of key nodes to determine the selected number of nodes in the first stage and threshold in the 2nd stage", required=False, default=-1) + parser.add_argument("--which_logs_folder", type=str, help="which_logs_folder", required=False, default="logs") # default="logs" + parser.add_argument("--log_subfolder", type=str, help="log_subfolder", required=False, default="demo_fit_native_add3") # default="demo_fit_native_add3" + # parser.add_argument("--use_ugraph", action="store_true") + parser.add_argument("--relative_dir_saved_mosca_model", type=str, help="relative_dir_saved_mosca_model. e.g.,", required=False) + args = parser.parse_args() + + seq_name=args.seq_name + dataset_name=args.dataset_name + depth_ratio=args.depth_ratio # 0.001 + threshold_min_set=args.threshold_min_set # 0.5 # paper-windmill 0.1 + version_key_edges=args.version_key_edges + extra_save_str=args.extra_save_str + ratio_key=args.ratio_key + # save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}/' # original + save_sub_folder=f'dr{depth_ratio}_thr{threshold_min_set}_v{version_key_edges}{extra_save_str}/' + logs_folder=args.which_logs_folder + relative_dir_saved_mosca_model=args.relative_dir_saved_mosca_model + log_subfolder=args.log_subfolder + if dataset_name=="iphone": + work_dir=f"/data/dataset/iphone_mosca_trained/{seq_name}/{logs_folder}/iphone_fit_native_add3" + save_dir=f"/data/dataset/iphone_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/iphone_mosca_trained/{seq_name}/" + voxel_size=0.2 + n_min_effective_frames=5 + # version_key_edges="max_contrib" + elif dataset_name=="davis": + work_dir=f"/data/dataset/custom_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # version_key_edges="minmax_distance_contrib" + elif dataset_name=="nvidia": + work_dir=f"/data/dataset/nvidia_mosca_trained/{seq_name}/{logs_folder}/nvidia_fit_native_add3" + save_dir=f"/data/dataset/nvidia_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/nvidia_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=3 + elif dataset_name=="diffusion4d": + work_dir=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/diffusion4d_mosca_graph_model/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/diffusion4d_mosca_trained/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + # elif dataset_name=="objaverse": # old + # group_name="000-001" + # # /data/dataset/objaverse_mosca_trained/000-001/3ae0b4e6a75a41d38a2adf8db048a703/000/mono/logs/demo_fit_native_add3 + # work_dir=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/{logs_folder}/demo_fit_native_add3" + # save_dir=f"/data/dataset/objaverse_mosca_graph_model/{group_name}/{seq_name}/000/mono/"+save_sub_folder + # ws=f"/data/dataset/objaverse_mosca_trained/{group_name}/{seq_name}/000/mono/" + # voxel_size=0.05 + # n_min_effective_frames=5 + elif dataset_name=="objaverse": + work_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/{logs_folder}/{log_subfolder}" # TODO demo_fit_native_add3_0 + save_dir=f"/data/dataset/objaverseBG_360_3deg_mosca_graph_model/{seq_name}/mono/"+save_sub_folder + ws=f"/data/dataset/objaverseBG_360_3deg_mosca_trained/{seq_name}/mono/" + voxel_size=0.1 # TODO + n_min_effective_frames=20 # TODO + elif dataset_name=="davis_mask": + work_dir=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/{logs_folder}/demo_fit_native_add3" + save_dir=f"/data/dataset/custom_mosca_graph_model_masked_4x/{seq_name}/"+save_sub_folder + ws=f"/data/dataset/custom_mosca_trained_masked_4x/{seq_name}/" + voxel_size=0.05 + n_min_effective_frames=5 + else: + raise ValueError("dataset_name not supported") + + backup_code(save_dir) + + file_name_graph=f"v12_parameters9mee.pth" # ee error propagation on both key and nonkey Gaussian #8 auto depth uncertainty + + if args.func_name=="main1": + main1(dataset_name,work_dir,save_dir,ws,relative_dir_saved_mosca_model) + elif args.func_name=="main23": + if args.depth_ratio is None or args.threshold_min_set is None or args.version_key_edges is None: + parser.error("--depth_ratio, --threshold_min_set, and --version_key_edges are required when func_name is 'main23'") + file_effective=main2(save_dir,voxel_size=voxel_size,n_min_effective_frames=n_min_effective_frames,ratio_key=ratio_key) + main3(save_dir,file_effective,file_name_graph,depth_ratio,version_key_edges=version_key_edges,threshold_min_set=threshold_min_set) #"max_contrib" minmax_distance_contrib + + + +if __name__ == "__main__": + # test2() + # test4() + # DON'T FORGET TO CHANGE! + run_from_outside() \ No newline at end of file diff --git a/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh new file mode 100644 index 0000000000000000000000000000000000000000..fade29ad45ed691be51bd485f7dd3274bb62fad1 --- /dev/null +++ b/000-044-041bc4c509634e52aa02d0bff5bf1ad8-000/mono/dr1.0_thr0.01_vmax_contrib/src_backup/run_preprocessing.sh @@ -0,0 +1,159 @@ +#!/bin/bash + +gpu_id=3 +dataset_name=iphone + +# 'backpack_' 'haru-sit_' 'paper-windmill' +seq_names=('paper-windmill' 'block' 'teddy' 'wheel' 'apple' 'spin' 'space-out') +# seq_names=('teddy' 'wheel') +# seq_names=('spin' 'space-out') +# seq_names=('paper-windmill') +# seq_names=('block') +# seq_names=('haru-sit_' 'handwavy_' 'mochi-high-five_' 'sriracha-tree_' 'creeper_' 'pillow_') +# seq_names=('haru-sit_') # 'backpack_' + +depth_ratio=0.1 +# threshold_min_set=0.1 # 0.1 paper-windmill # 0.5 others +threshold_min_set_dict=( + ["paper-windmill"]=0.1 + ["teddy"]=0.5 + ["wheel"]=0.5 + ["apple"]=0.5 + ["spin"]=0.5 + ["space-out"]=0.5 + ["haru-sit_"]=0.5 + ["handwavy_"]=0.5 + ["mochi-high-five_"]=0.5 + ["sriracha-tree_"]=0.5 + ["creeper_"]=0.5 + ["pillow_"]=0.5 +) +version_key_edges=max_contrib # max_contrib minmax_distance_contrib max_native +ratio_key=0.02 # default -1 +extra_save_str=_ab_key_ratio_$ratio_key #2 + +# Loop and print each name +for seq_name in "${seq_names[@]}" +do + echo "${seq_name}" +done + +# Loop over each sequence +for seq_name in "${seq_names[@]}" +do + echo "Starting training for ${seq_name}" + + threshold_min_set=${threshold_min_set_dict[$seq_name]} + + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main1 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str # comment if no extra_save_str + CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ + --seq_name $seq_name \ + --func_name main23 \ + --dataset_name $dataset_name \ + --depth_ratio $depth_ratio \ + --threshold_min_set $threshold_min_set \ + --version_key_edges $version_key_edges \ + --extra_save_str $extra_save_str \ + --ratio_key $ratio_key # ablation study + + echo "Finished training for ${seq_name}" + echo "*****************************************" + echo "*****************************************" +done + + + + + + +# gpu_id=2 +# dataset_name=davis +# seq_names=('camel' 'car-turn' 'drift-straight' 'helicopter' 'mbike-santa' 'koala' 'parkour' 'sheep' 'soccerball' 'schoolgirls') # 'bear' 'rhino' +# # seq_names=('camel' 'train' 'car-roundabout' 'car-turn' 'bike-packing' 'drift-straight' 'helicopter' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'mbike-santa' 'parkour' 'pigs' 'rhino' 'schoolgirls' 'sheep' 'soccerball' 'swing' 'bear') +# # seq_names=('train' 'car-roundabout' 'bike-packing' 'breakdance' 'breakdance-flare' 'crossing' 'dance-twirl' 'koala' 'parkour' 'pigs' 'schoolgirls' 'sheep' 'soccerball' 'swing') + +# depth_ratio=0.001 # 0.1 +# threshold_min_set=0.1 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done + + + +# gpu_id=1 +# dataset_name=nvidia + +# # seq_names=('Balloon1' 'Balloon2' 'Jumping' 'Playground' 'Skating' 'Truck' 'Umbrella') # +# # seq_names=('Balloon1' 'Jumping' 'Truck' ) # 'Umbrella' +# seq_names=('Balloon2' 'Playground' 'Skating') + +# depth_ratio=0.1 +# threshold_min_set=0.01 # 0.5 +# version_key_edges=minmax_distance_contrib # max_contrib minmax_distance_contrib + +# # Loop and print each name +# for seq_name in "${seq_names[@]}" +# do +# echo "${seq_name}" +# done + +# # Loop over each sequence +# for seq_name in "${seq_names[@]}" +# do +# echo "Starting training for ${seq_name}" + +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main1 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges +# CUDA_VISIBLE_DEVICES=$gpu_id python graph_model_preprocessing5.py \ +# --seq_name $seq_name \ +# --func_name main23 \ +# --dataset_name $dataset_name \ +# --depth_ratio $depth_ratio \ +# --threshold_min_set $threshold_min_set \ +# --version_key_edges $version_key_edges + +# echo "Finished training for ${seq_name}" +# echo "*****************************************" +# echo "*****************************************" +# done \ No newline at end of file