Add FoundationPose training metadata, previews, and transfer guide

.gitattributes

@@ -64,3 +64,5 @@ previews/ycbv/test-geometry.glb filter=lfs diff=lfs merge=lfs -text
 previews/sage/test-geometry.glb filter=lfs diff=lfs merge=lfs -text
 previews/sage/train-geometry.glb filter=lfs diff=lfs merge=lfs -text
 precomputed_slat/previews/3dfront/fffeddd2444aac00cee8/object_latent_cells.glb filter=lfs diff=lfs merge=lfs -text
+previews/foundationpose/test-occupancy.glb filter=lfs diff=lfs merge=lfs -text
+previews/foundationpose/train-occupancy.glb filter=lfs diff=lfs merge=lfs -text

README.md

@@ -24,14 +24,20 @@ configs:
   data_files:
   - split: test
     path: catalog/ycbv/test.jsonl
+- config_name: foundationpose
+  data_files:
+  - split: test
+    path: catalog/foundationpose/test.jsonl
+  - split: train
+    path: catalog/foundationpose/train.jsonl
 ---
 
 # PartProbe3D preprocessed datasets
 
-Saved 64³ training and evaluation inputs for PartProbe3D. These archives preserve
-RGB, visible object masks, depth and validity, camera calibration, posed meshes,
-scene occupancy and independent per-object voxel occupancy. Targets describe mesh
-triangle surfaces, including hidden surfaces and overlapping object channels.
+Saved 64³ training and evaluation inputs for PartProbe3D preserve calibrated observations and voxel targets.
+The 3D-FRONT, SAGE-10k, and YCB-V bundles also retain posed meshes.
+FoundationPose contains an export for training without meshes.
+Targets describe triangle surfaces, including hidden surfaces and overlapping object channels.
 
 | Dataset | Split | Views | Scenes |
 | --- | --- | ---: | ---: |
@@ -40,6 +46,8 @@ triangle surfaces, including hidden surfaces and overlapping object channels.
 | 3dfront | test | 10,066 | 935 |
 | 3dfront | train | 39,930 | 3,703 |
 | ycbv | test | 900 | 900 |
+| foundationpose | train | 42,256 | 42,256 |
+| foundationpose | test | 4,692 | 4,692 |
 
 3D-FRONT uses the pinned last 1,000 MIDI rooms for testing and all other rooms
 for training; all views of a room stay together. There is no validation split.
@@ -61,8 +69,126 @@ The completed extent dataset retains 19,408 training/evaluation views;
 
 Archive inventories and SHA-256 checksums are in `bundles/<dataset>/index.json`.
 The extent archives exclude diagnostic subsets, raw downloads, model weights, and model-specific feature caches.
-FoundationPose currently has raw archives only.
+FoundationPose archive publication is in progress. The complete inventory appears after verification of every remote archive.
+
+
+## FoundationPose on another server
+
+The FoundationPose bundle preserves original RGB, depth, visible masks, cameras, object mapping, and complete per-object voxel occupancy at 64³.
+Its source manifest has SHA-256 `2c4dcd1cd1e2e56a0eb08f2f92da320f299729c8ce7a2f52828c272cbc1fa0ad`.
+The extracted files occupy 87.82 GB before filesystem overhead.
+The export excludes raw archives, meshes, model weights, image features, SLAT latents, and GT occlusion relations.
+
+The loaders resolve input paths relative to the restored dataset.
+Historical absolute paths remain provenance entries.
+SS preparation creates the union of complete occupancy for objects with supplied masks and excludes other object voxels.
+
+### Restore the data
+
+Use Python 3.12 and a compatible PyTorch/CUDA installation.
+Run these commands from the repository root.
+
+1. Install the project dependencies.
+
+   ```bash
+   python -m pip install -e '.[preprocess,ss-multiobj-depth,tracking]'
+   ```
+
+2. Restore the prepared observations and voxel targets.
+
+   ```bash
+   python -m preprocess.huggingface download \
+     --repo cubec/PartProbe3D-preprocessed --dataset foundationpose --cleanup
+   ```
+
+3. Initialize the pinned Pixal3D source.
+
+   ```bash
+   git submodule update --init third_party/Pixal3D
+   ```
+
+`--cleanup` retains extracted data and removes each verified temporary archive.
+Allow at least 100 GB for data restoration, plus space for weights and model features.
+Use `--revision <Hub commit>` to retain one dataset snapshot.
+
+### Acquire SS weights
+
+All required weights for the following SS configuration are available from Hugging Face.
+The commands pin revisions with the same configuration and weight bytes as the verified local files.
+DINOv3 requires approved access to its [gated model repository](https://huggingface.co/facebook/dinov3-vitl16-pretrain-lvd1689m).
+
+1. Authenticate with an account that can download DINOv3.
+
+   ```bash
+   hf auth login
+   ```
+
+2. Download the native SS flow and decoder from [Pixal3D](https://huggingface.co/TencentARC/Pixal3D).
+
+   ```bash
+   hf download TencentARC/Pixal3D \
+     --revision b0cb2e1b794cab9aa0ac38a95d794a4d9337437f \
+     --local-dir checkpoints/Pixal3D \
+     --include 'ckpts/ss_flow_img_dit_1_3B_64_bf16.*' 'ckpts/ss_dec_conv3d_16l8_fp16.*'
+   ```
+
+3. Download the SS encoder and paired decoder from [TRELLIS](https://huggingface.co/microsoft/TRELLIS-image-large).
+
+   ```bash
+   hf download microsoft/TRELLIS-image-large \
+     --revision 25e0d31ffbebe4b5a97464dd851910efc3002d96 \
+     --local-dir checkpoints/TRELLIS \
+     --include 'ckpts/ss_enc_conv3d_16l8_fp16.*' 'ckpts/ss_dec_conv3d_16l8_fp16.safetensors'
+   ```
+
+4. Download DINOv3.
+
+   ```bash
+   hf download facebook/dinov3-vitl16-pretrain-lvd1689m \
+     --revision ea8dc2863c51be0a264bab82070e3e8836b02d51 \
+     --local-dir checkpoints/DINOv3 --include config.json model.safetensors
+   ```
+
+### Prepare features and train SS
+
+The [FoundationPose configuration](https://github.com/cube-c/PartProbe3D/blob/slat-shape/ss_multiobj_depth/configs/foundationpose.json) uses GT depth, `depth_embedding` voxel conditioning, global cross-attention, and rank-8 NoRA.
+It creates features under `outputs/ss_multiobj_depth/foundationpose/features` and preserves the saved training/test splits.
+It starts a new run with W&B online by default.
+
+1. Prepare local features and encoded SS targets.
+
+   ```bash
+   python -m ss_multiobj_depth.prepare --config ss_multiobj_depth/configs/foundationpose.json
+   ```
+
+2. Start training.
+
+   ```bash
+   python -m ss_multiobj_depth.train --config ss_multiobj_depth/configs/foundationpose.json
+   ```
+
+This configuration needs no source archives, meshes, or additional dataset downloads.
+The `occlusion_object` variant additionally requires a matching GT occlusion cache.
+That cache requires mesh geometry during preparation and is absent from this FoundationPose bundle.
+New SLAT shape encoding and mesh-based evaluation also require geometry or separately transferred targets.
+Exact resume of an existing run requires its matching checkpoint, feature manifests, source hashes, and settings.
+
+
+### FoundationPose previews
+
+These previews show saved inputs and GT voxel targets. They use no learned checkpoint.
+
+![FoundationPose train RGB](previews/foundationpose/train-rgb.png)
+
+![FoundationPose train masks](previews/foundationpose/train-overlay.png)
+
+[GT occupancy, train (GLB)](previews/foundationpose/train-occupancy.glb)
+
+![FoundationPose test RGB](previews/foundationpose/test-rgb.png)
+
+![FoundationPose test masks](previews/foundationpose/test-overlay.png)
 
+[GT occupancy, test (GLB)](previews/foundationpose/test-occupancy.glb)
 
 ## GT occlusion relations
 

catalog.json

@@ -60,6 +60,31 @@
       "uploaded_bytes": 3862999040,
       "uploaded_archives": 5,
       "uploaded": true
+    },
+    {
+      "name": "foundationpose",
+      "source_complete": true,
+      "artifact_kind": "extent-training",
+      "geometry_included": false,
+      "feature_caches_included": false,
+      "occlusion_included": false,
+      "views": 46948,
+      "counts": {
+        "test": {
+          "views": 4692,
+          "scenes": 4692
+        },
+        "train": {
+          "views": 42256,
+          "scenes": 42256
+        }
+      },
+      "source_manifest_sha256": "2c4dcd1cd1e2e56a0eb08f2f92da320f299729c8ce7a2f52828c272cbc1fa0ad",
+      "packaged": false,
+      "bytes": 33531084800,
+      "uploaded_bytes": 68556800,
+      "uploaded_archives": 1,
+      "uploaded": false
     }
   ]
 }

metadata/foundationpose/conversion_policy.json

@@ -0,0 +1,42 @@
+{
+  "schema": "partprobe3d.views",
+  "capture_manifest_sha256": "17551a7595f369a2c1341304119f01a740c0250ac265f7fddb40148c3ea1d49c",
+  "settings": {
+    "resolution": 64,
+    "padding": 1.0,
+    "center": null,
+    "cube_size": null,
+    "minimum_mask_pixels": 64,
+    "minimum_objects": 2,
+    "max_voxels": 0,
+    "extent_strategy": "point_cloud",
+    "extent_spec": {
+      "strategy": "point_cloud",
+      "options": {
+        "near_trim_fraction": 0.0525,
+        "far_trim_fraction": 0.0525,
+        "far_margin": 0.125,
+        "padding": 1.31
+      },
+      "parameters": {},
+      "mode": "train"
+    }
+  },
+  "generator_sha256": "0987010ec3475e5e09823497232cb26fac315fef42499943acd9912ae17f267a",
+  "adapter_sha256": "0513dcab76dc7f5e07bf36e4e5641869211ab0605afd60387c1d575668f6a8ac",
+  "exporter_sha256": "3d84f7e02ee746e7c010086e380e120c070ca69e4442284236aae4967c54e770",
+  "camera_sha256": "3670c55189445fe07e0fec3eb57981655e9359794e186a9244fbe4251db4c674",
+  "storage_sha256": "18c2240fd72150c9fea9a13b0ddef9cd307b7aba5deafd876c9f43ddade39428",
+  "dataset_sha256": "becdacb3842a0a7a8e52007897853125b834b45b87cd59e729f4477b6ea48351",
+  "voxelizer_sha256": "0504d234113b42534081742b73a4ab3ca3c1df7c6a338c44be3800597c910fe5",
+  "extent_cube_sha256": "c9f7b0e24940c1c65f9398750a55e4ac730e93b784985a5583770cf94db9679f",
+  "extent_base_sha256": "3202d6f613b9766593b9a215d86e8ccd678f253d2e7c815a6bb911c2bdc253fb",
+  "ss_depth_extent_sha256": "1a7bf1d6fb62a052507706af793df3206bcb28218fbbdb6d71ee2c9a3c3ddfd6",
+  "pixal3d_extent_sha256": "e8d8d5e77511e2847f23df3c7b9ba3d7da0b18af0481378765320ff726960c9f",
+  "point_cloud_extent_sha256": "114d4a0258f8bb9d37c5ffabcab4090dd6222c66b4d8747ddbb6da97c676d278",
+  "extent_factory_sha256": "e18dc4d8f9f1ef671bc537cae51877f2242bbe0a3bf513ba6a1e5f41dd186c09",
+  "extent_specification_sha256": "b0e8f0a7809192dbbfcb9da58108a632448ece4266630abafbd4571625dd4fa5",
+  "surface_voxelizer_sha256": "298547e642b9b85058d78f4bb390e1a0950c4f0d6ccb52f49207786badddd4b9",
+  "surface_alignment_sha256": "3860ad05627596a323916460fb8dab83a5ee6267007cf7f2adf9223f4d3b8971",
+  "geometry_reader_sha256": "3c0d2a0524ae06ff6a8a9c831b45cd82967f690202189ad843b87e8c9818ea0a"
+}

metadata/foundationpose/model_dependencies.json

@@ -0,0 +1,71 @@
+[
+  {
+    "repo": "TencentARC/Pixal3D",
+    "revision": "b0cb2e1b794cab9aa0ac38a95d794a4d9337437f",
+    "local_root": "checkpoints/Pixal3D",
+    "gated": false,
+    "files": [
+      {
+        "file": "ckpts/ss_flow_img_dit_1_3B_64_bf16.json",
+        "bytes": 503,
+        "sha256": "2b3ef4c63b72a3de8adf745bd9d9d36b7960bdd7ba3740a6fa4d18fea6c475dd"
+      },
+      {
+        "file": "ckpts/ss_flow_img_dit_1_3B_64_bf16.safetensors",
+        "bytes": 5359822584,
+        "sha256": "6d0147bb347eb61fa81c304cc68c7fd69d447e91870357a549b47e1d0fd77242"
+      },
+      {
+        "file": "ckpts/ss_dec_conv3d_16l8_fp16.json",
+        "bytes": 245,
+        "sha256": "646781293f1cda74720de85d1cef50a957fb4aebd9a4bd014e454e32f2330ac5"
+      },
+      {
+        "file": "ckpts/ss_dec_conv3d_16l8_fp16.safetensors",
+        "bytes": 147591972,
+        "sha256": "1c76d4a40519aa2d711cc263a8404105231ac26db31d946bed48b84fee79009a"
+      }
+    ]
+  },
+  {
+    "repo": "microsoft/TRELLIS-image-large",
+    "revision": "25e0d31ffbebe4b5a97464dd851910efc3002d96",
+    "local_root": "checkpoints/TRELLIS",
+    "gated": false,
+    "files": [
+      {
+        "file": "ckpts/ss_enc_conv3d_16l8_fp16.json",
+        "bytes": 244,
+        "sha256": "12efe92a0d7dcd790f251acb94a6950957ea4398268e2838ef7319c3f20b071e"
+      },
+      {
+        "file": "ckpts/ss_enc_conv3d_16l8_fp16.safetensors",
+        "bytes": 119068016,
+        "sha256": "107874eeaa0feb82f51b19db5da7db534fb7e7f19e5a122b9ff1bc2e258bfc6d"
+      },
+      {
+        "file": "ckpts/ss_dec_conv3d_16l8_fp16.safetensors",
+        "bytes": 147591972,
+        "sha256": "1c76d4a40519aa2d711cc263a8404105231ac26db31d946bed48b84fee79009a"
+      }
+    ]
+  },
+  {
+    "repo": "facebook/dinov3-vitl16-pretrain-lvd1689m",
+    "revision": "ea8dc2863c51be0a264bab82070e3e8836b02d51",
+    "local_root": "checkpoints/DINOv3",
+    "gated": "manual",
+    "files": [
+      {
+        "file": "config.json",
+        "bytes": 745,
+        "sha256": "135ecd23e34a70b6fbed8b083fdecb319b7e3a54e3d849258bbe4ddcf1783bb5"
+      },
+      {
+        "file": "model.safetensors",
+        "bytes": 1212559808,
+        "sha256": "dcb2e45127cccbf1601e5f42fef165eea275c8e5213197e8dcf3f48822718179"
+      }
+    ]
+  }
+]

metadata/foundationpose/previews.json

@@ -0,0 +1,23 @@
+{
+  "checkpoint": null,
+  "source_manifest_sha256": "2c4dcd1cd1e2e56a0eb08f2f92da320f299729c8ce7a2f52828c272cbc1fa0ad",
+  "target": "Union of complete per-object voxel occupancy for prompted objects only",
+  "views": [
+    {
+      "id": "02a27246d608ce3981f9",
+      "split": "train",
+      "prompted_objects": 7,
+      "prompted_union_voxels": 14246,
+      "source_view_stamp_sha256": "0fb108e2c374fb0868de7106e8b61db9a6ac7aa564816ba31e4bb392928e4999",
+      "source_grid_sha256": "6c569885bbc2dbf82433cc1a3dca637dc1700680ecca9532fabc32100aed2f6f"
+    },
+    {
+      "id": "0f5bc64d32b8fce7f2f0",
+      "split": "test",
+      "prompted_objects": 7,
+      "prompted_union_voxels": 10795,
+      "source_view_stamp_sha256": "15cbcde8c4f1b37788dbab4aafe8eae6d94b4fd5ae365369b3dbf206d8f65379",
+      "source_grid_sha256": "4d0587dd5745b983958987936fa20e0670618f914eb1296e2e18fe89e8808b2f"
+    }
+  ]
+}

metadata/foundationpose/ss_config.json

@@ -0,0 +1,59 @@
+{
+  "schema": "ss-multiobj-depth-v1",
+  "dataset": "datasets/foundationpose/extent/point_cloud",
+  "cache": "outputs/ss_multiobj_depth/foundationpose/features",
+  "output": "outputs/ss_multiobj_depth/foundationpose/train",
+  "checkpoints": "checkpoints/training/ss_multiobj_depth/foundationpose",
+  "device": "cuda:0",
+  "seed": 0,
+  "depth_source": "gt",
+  "moge_config": "ss_multiobj_depth/configs/moge3.json",
+  "models": {
+    "upstream": "third_party/Pixal3D",
+    "upstream_revision": "cdbb2bbffbf4e6f298b5f2af3d1d76a8d823d2af",
+    "dino": "checkpoints/DINOv3",
+    "encoder": "checkpoints/TRELLIS/ckpts/ss_enc_conv3d_16l8_fp16",
+    "paired_decoder": "checkpoints/TRELLIS/ckpts/ss_dec_conv3d_16l8_fp16",
+    "decoder": "checkpoints/Pixal3D/ckpts/ss_dec_conv3d_16l8_fp16",
+    "flow": "checkpoints/Pixal3D/ckpts/ss_flow_img_dit_1_3B_64_bf16"
+  },
+  "model": {
+    "cross_attention_mode": "global",
+    "depth_channels": 128,
+    "rank": 8,
+    "alpha": 8.0,
+    "voxel_condition_mode": "depth_embedding",
+    "voxel_condition_hidden_channels": 64
+  },
+  "training": {
+    "steps": 20000,
+    "batch_size": 2,
+    "accumulation": 4,
+    "workers": 4,
+    "learning_rate": 3e-05,
+    "depth_learning_rate": 0.0001,
+    "warmup": 250,
+    "cfg_dropout": 0.1,
+    "depth_augmentation": {
+      "noise_probability": 0.3,
+      "noise_std_fraction": 0.015625,
+      "pixel_dropout": 0.1
+    },
+    "save_every": 500,
+    "eval_every": 500,
+    "eval_limit": 8,
+    "wandb_mode": "online",
+    "wandb_project": "PartProbe3D",
+    "wandb_entity": null
+  },
+  "sampling": {
+    "steps": 12,
+    "cfg_strength": 7.5,
+    "cfg_rescale": 0.7,
+    "cfg_interval": [
+      0.6,
+      1.0
+    ],
+    "rescale_t": 5
+  }
+}

metadata/foundationpose/training_export.json

@@ -0,0 +1,12 @@
+{
+  "schema": "partprobe3d.training-export.v1",
+  "complete": true,
+  "source_manifest_sha256": "2c4dcd1cd1e2e56a0eb08f2f92da320f299729c8ce7a2f52828c272cbc1fa0ad",
+  "dataset": "FoundationPose",
+  "views": 46948,
+  "geometry_included": false,
+  "feature_caches_included": false,
+  "files": 1919211,
+  "logical_bytes": 87820343901,
+  "occlusion_included": false
+}

metadata/foundationpose/validation_skipped.json

@@ -0,0 +1,53 @@
+{
+  "complete": true,
+  "full_validation": false,
+  "scope": "sampled",
+  "full_validation_skipped": true,
+  "dataset": "FoundationPose",
+  "dataset_views": 46948,
+  "validated_views": 1000,
+  "invalidated_views": 0,
+  "manifest_sha256": "2c4dcd1cd1e2e56a0eb08f2f92da320f299729c8ce7a2f52828c272cbc1fa0ad",
+  "request_sha256": "e093cdc2c60db19686d5f342aa3c2d58d8261b71b98275e658649af004cc8f99",
+  "reason": "The user requested skipping full validation for this run after all 1,000 sampled views passed.",
+  "sample_audit_sha256": "c1ed1c0d7e50758daf88f8a4926e6462347dae9e67892387c9b0ffcf7b305fc3",
+  "sample_source_manifest_sha256": "1de185303d1debe4199d0a6a971a77192d2b538949b5e0704534ad0b4de582b9",
+  "sample_validator_sha256": "ee5d012b440133fb95407579b9ac64890da520800fc802a37fd5dbd710a4da78",
+  "sample_validation": {
+    "complete": true,
+    "schema": "partprobe3d.views",
+    "dataset": "FoundationPose",
+    "manifest_sha256": "85bae223753c0fdb4d9f6b1c1458f7dee14230120ecb8c22fef46be405448197",
+    "validated_views": 1000,
+    "validated_scenes": 1000,
+    "validated_grids": 1000,
+    "binary_masks": 31409,
+    "counts": {
+      "test": {
+        "views": 94,
+        "scenes": 94
+      },
+      "train": {
+        "views": 906,
+        "scenes": 906
+      }
+    },
+    "overlapping_voxel_rows_across_views": 280072,
+    "unprompted_only_voxel_rows_across_views": 2116873,
+    "checks": [
+      "artifact hashes",
+      "RGB/depth/label alignment",
+      "binary mask/object identity",
+      "sparse/dense occupancy equality",
+      "shared scene GT and prompt columns",
+      "scene split separation",
+      "unique cache filenames"
+    ]
+  },
+  "global_checks": [
+    "complete manifest",
+    "scene split separation",
+    "unique cache filenames"
+  ],
+  "export_checks": "Training export retains its file checksums and portability checks."
+}

previews/foundationpose/test-occupancy.glb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dea9f3c85096b53355b239970efac3f957a2954e84bb748596fd52c28aba469
+size 2938436

previews/foundationpose/train-occupancy.glb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e09ce1314836251b91c0ce9af673a340e634cb36a099a1b5b4757be6d11dd38
+size 3877112