diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-bSqkOH3Y80_00013_0_95.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-bSqkOH3Y80_00013_0_95.json new file mode 100644 index 0000000000000000000000000000000000000000..f4ba80c76172739ec00f9327fc2e7df2c7f88f5a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-bSqkOH3Y80_00013_0_95.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_-bSqkOH3Y80_00013_0_95", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/-bSqkOH3Y80_00013_0_95", + "duration": 4.0 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/-bSqkOH3Y80_00013_0_95_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/-bSqkOH3Y80_00013_0_95", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-bgPTgJJlv4_00019_8_72.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-bgPTgJJlv4_00019_8_72.json new file mode 100644 index 0000000000000000000000000000000000000000..8a301d37b292583bcf6d9aac9388a873559056e1 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-bgPTgJJlv4_00019_8_72.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_-bgPTgJJlv4_00019_8_72", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/-bgPTgJJlv4_00019_8_72", + "duration": 2.169 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/-bgPTgJJlv4_00019_8_72_joint0_skel.jpg", + "time": 0.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/-bgPTgJJlv4_00019_8_72", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-gxw2PyiUJA_00045_573_723.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-gxw2PyiUJA_00045_573_723.json new file mode 100644 index 0000000000000000000000000000000000000000..d1bceb63effd667acc0612e03bb7bbe0eb93eabd --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-gxw2PyiUJA_00045_573_723.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_-gxw2PyiUJA_00045_573_723", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/-gxw2PyiUJA_00045_573_723", + "duration": 5.072 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is walking across a grassy field from a deflated blue and white parachute toward the camera to stop and speak while gesturing.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/-gxw2PyiUJA_00045_573_723_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/-gxw2PyiUJA_00045_573_723", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-ne-ss1e51c_00005_247_387.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-ne-ss1e51c_00005_247_387.json new file mode 100644 index 0000000000000000000000000000000000000000..8eb03694bc0be70936ff6f77332c66e4947b0413 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_-ne-ss1e51c_00005_247_387.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_-ne-ss1e51c_00005_247_387", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/-ne-ss1e51c_00005_247_387", + "duration": 4.739 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/-ne-ss1e51c_00005_247_387_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/-ne-ss1e51c_00005_247_387", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0-NOMvTGx3Q_00034_0_80.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0-NOMvTGx3Q_00034_0_80.json new file mode 100644 index 0000000000000000000000000000000000000000..cae8f19dfa242c7624b8f7a5124170ef400352e7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0-NOMvTGx3Q_00034_0_80.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_0-NOMvTGx3Q_00034_0_80", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/0-NOMvTGx3Q_00034_0_80", + "duration": 2.7 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0-NOMvTGx3Q_00034_0_80_joint0_skel.jpg", + "time": 0.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0-NOMvTGx3Q_00034_0_80_joint1_skel.jpg", + "time": 1.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/0-NOMvTGx3Q_00034_0_80", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_003q7lpJFGI_00003_934_1115.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_003q7lpJFGI_00003_934_1115.json new file mode 100644 index 0000000000000000000000000000000000000000..d0225a435efc9afdfa20152b7f2e76a249555b17 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_003q7lpJFGI_00003_934_1115.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_003q7lpJFGI_00003_934_1115", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/003q7lpJFGI_00003_934_1115", + "duration": 3.104 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/003q7lpJFGI_00003_934_1115_joint0_skel.jpg", + "time": 0.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/003q7lpJFGI_00003_934_1115", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0CyBZEykTu4_00001_59_258.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0CyBZEykTu4_00001_59_258.json new file mode 100644 index 0000000000000000000000000000000000000000..c995fb71befcb0acf6eac025e51e3b57b15255d7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0CyBZEykTu4_00001_59_258.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_0CyBZEykTu4_00001_59_258", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/0CyBZEykTu4_00001_59_258", + "duration": 6.674 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0CyBZEykTu4_00001_59_258_joint0_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0CyBZEykTu4_00001_59_258_joint1_skel.jpg", + "time": 3.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/0CyBZEykTu4_00001_59_258", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0FIuy2pklxQ_00025_0_105.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0FIuy2pklxQ_00025_0_105.json new file mode 100644 index 0000000000000000000000000000000000000000..b88600682e35c20cd1a54e4b3f493e69ca4d285a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0FIuy2pklxQ_00025_0_105.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_0FIuy2pklxQ_00025_0_105", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/0FIuy2pklxQ_00025_0_105", + "duration": 4.375 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0FIuy2pklxQ_00025_0_105_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/0FIuy2pklxQ_00025_0_105", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0Oxd5OiY258_00079_69_183.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0Oxd5OiY258_00079_69_183.json new file mode 100644 index 0000000000000000000000000000000000000000..98f871a525a521965d77194576a4e18456276795 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0Oxd5OiY258_00079_69_183.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_0Oxd5OiY258_00079_69_183", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/0Oxd5OiY258_00079_69_183", + "duration": 2.3 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0Oxd5OiY258_00079_69_183_joint0_skel.jpg", + "time": 0.3 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/0Oxd5OiY258_00079_69_183", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0TVSPT_n_xk_00044_23_98.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0TVSPT_n_xk_00044_23_98.json new file mode 100644 index 0000000000000000000000000000000000000000..4215cabc6b48d3098ebba8464ffcb4ce57774c3a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0TVSPT_n_xk_00044_23_98.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_0TVSPT_n_xk_00044_23_98", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/0TVSPT_n_xk_00044_23_98", + "duration": 3.13 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is slowly pivoting in a cold, blue-lit environment while remaining stationary below and then subtly shifting to show attentiveness toward an external event.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0TVSPT_n_xk_00044_23_98_joint0_skel.jpg", + "time": 2.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/0TVSPT_n_xk_00044_23_98", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0XeuCK3PfqA_00024_87_316.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0XeuCK3PfqA_00024_87_316.json new file mode 100644 index 0000000000000000000000000000000000000000..a859b7ab04124f7ff2f5787f522adea7400c2e60 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0XeuCK3PfqA_00024_87_316.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_0XeuCK3PfqA_00024_87_316", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/0XeuCK3PfqA_00024_87_316", + "duration": 7.675 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0XeuCK3PfqA_00024_87_316_joint0_skel.jpg", + "time": 3.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/0XeuCK3PfqA_00024_87_316", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0h7AlrKLslw_00056_0_313.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0h7AlrKLslw_00056_0_313.json new file mode 100644 index 0000000000000000000000000000000000000000..60c4373477651b8b4fa35e32144e75cecbbe4099 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_0h7AlrKLslw_00056_0_313.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_0h7AlrKLslw_00056_0_313", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/0h7AlrKLslw_00056_0_313", + "duration": 5.267 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0h7AlrKLslw_00056_0_313_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0h7AlrKLslw_00056_0_313_joint1_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/0h7AlrKLslw_00056_0_313_joint2_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/0h7AlrKLslw_00056_0_313", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CL4WdaOqcz4_00000_1019_1180.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CL4WdaOqcz4_00000_1019_1180.json new file mode 100644 index 0000000000000000000000000000000000000000..40a54ab58409cb4a24d83c4861837486c129e315 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CL4WdaOqcz4_00000_1019_1180.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_CL4WdaOqcz4_00000_1019_1180", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/CL4WdaOqcz4_00000_1019_1180", + "duration": 5.539 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/CL4WdaOqcz4_00000_1019_1180_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/CL4WdaOqcz4_00000_1019_1180_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/CL4WdaOqcz4_00000_1019_1180", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CivzEXLprZw_00005_835_2081.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CivzEXLprZw_00005_835_2081.json new file mode 100644 index 0000000000000000000000000000000000000000..19468f6b8f8f36cad31fe42ffdaa3f148f7c1fa5 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CivzEXLprZw_00005_835_2081.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_CivzEXLprZw_00005_835_2081", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/CivzEXLprZw_00005_835_2081", + "duration": 20.784 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/CivzEXLprZw_00005_835_2081_joint0_skel.jpg", + "time": 9.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/CivzEXLprZw_00005_835_2081", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Cn2s3-690wI_00004_105_199.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Cn2s3-690wI_00004_105_199.json new file mode 100644 index 0000000000000000000000000000000000000000..12b77118d57b82221f0661335bd3e1a458b6c1aa --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Cn2s3-690wI_00004_105_199.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_Cn2s3-690wI_00004_105_199", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/Cn2s3-690wI_00004_105_199", + "duration": 3.17 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is presenting a weather forecast by indicating a specific region on the map and then returning to a neutral, centered orientation toward the viewer.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/Cn2s3-690wI_00004_105_199_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/Cn2s3-690wI_00004_105_199", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CwVGkPsOM9g_00024_34_140.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CwVGkPsOM9g_00024_34_140.json new file mode 100644 index 0000000000000000000000000000000000000000..76b23b22dfbb4bf20be45d3a50335b5e486dc330 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CwVGkPsOM9g_00024_34_140.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_CwVGkPsOM9g_00024_34_140", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/CwVGkPsOM9g_00024_34_140", + "duration": 3.571 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/CwVGkPsOM9g_00024_34_140_joint0_skel.jpg", + "time": 0.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/CwVGkPsOM9g_00024_34_140", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CwzjmP57_Jw_00008_1542_1744.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CwzjmP57_Jw_00008_1542_1744.json new file mode 100644 index 0000000000000000000000000000000000000000..f7c0aa310ad3af81a17c9a033e77a5a8732c6be7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_CwzjmP57_Jw_00008_1542_1744.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_CwzjmP57_Jw_00008_1542_1744", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/CwzjmP57_Jw_00008_1542_1744", + "duration": 3.4 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/CwzjmP57_Jw_00008_1542_1744_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/CwzjmP57_Jw_00008_1542_1744", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_D0-b-ovg4VY_00031_183_276.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_D0-b-ovg4VY_00031_183_276.json new file mode 100644 index 0000000000000000000000000000000000000000..6aaf806f7ea5ae2979060a0699bd8b0c9a1c113b --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_D0-b-ovg4VY_00031_183_276.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_D0-b-ovg4VY_00031_183_276", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/D0-b-ovg4VY_00031_183_276", + "duration": 3.963 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/D0-b-ovg4VY_00031_183_276_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/D0-b-ovg4VY_00031_183_276", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DKMRRpmgitc_00031_0_398.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DKMRRpmgitc_00031_0_398.json new file mode 100644 index 0000000000000000000000000000000000000000..5f1dafb63652a571e9651efef50203c98ad0d1a9 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DKMRRpmgitc_00031_0_398.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_DKMRRpmgitc_00031_0_398", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/DKMRRpmgitc_00031_0_398", + "duration": 13.292 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DKMRRpmgitc_00031_0_398_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DKMRRpmgitc_00031_0_398_joint1_skel.jpg", + "time": 6.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/DKMRRpmgitc_00031_0_398", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DQGAcsxCzPw_00002_47_121.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DQGAcsxCzPw_00002_47_121.json new file mode 100644 index 0000000000000000000000000000000000000000..40c2d34b60c760536153269b17f18b2cf6a54e29 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DQGAcsxCzPw_00002_47_121.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_DQGAcsxCzPw_00002_47_121", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/DQGAcsxCzPw_00002_47_121", + "duration": 2.536 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DQGAcsxCzPw_00002_47_121_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/DQGAcsxCzPw_00002_47_121", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DTJjOysP-bI_00000_314_397.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DTJjOysP-bI_00000_314_397.json new file mode 100644 index 0000000000000000000000000000000000000000..b09de0656790b45c46aa12684deecd49c18e8f9b --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DTJjOysP-bI_00000_314_397.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_DTJjOysP-bI_00000_314_397", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/DTJjOysP-bI_00000_314_397", + "duration": 2.834 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DTJjOysP-bI_00000_314_397_joint0_skel.jpg", + "time": 0.8 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DTJjOysP-bI_00000_314_397_joint1_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/DTJjOysP-bI_00000_314_397", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DWgCNBhHih8_00005_234_474.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DWgCNBhHih8_00005_234_474.json new file mode 100644 index 0000000000000000000000000000000000000000..1115713811dced091910ca1aded8d3375fff77e3 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DWgCNBhHih8_00005_234_474.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_DWgCNBhHih8_00005_234_474", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/DWgCNBhHih8_00005_234_474", + "duration": 9.603 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DWgCNBhHih8_00005_234_474_joint0_skel.jpg", + "time": 7.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/DWgCNBhHih8_00005_234_474", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DjYdSs-Uhsw_00032_0_96.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DjYdSs-Uhsw_00032_0_96.json new file mode 100644 index 0000000000000000000000000000000000000000..21ee9253a62b0b6023df25c6d85c6f2c6876554d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DjYdSs-Uhsw_00032_0_96.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_DjYdSs-Uhsw_00032_0_96", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/DjYdSs-Uhsw_00032_0_96", + "duration": 4.046 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DjYdSs-Uhsw_00032_0_96_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DjYdSs-Uhsw_00032_0_96_joint1_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/DjYdSs-Uhsw_00032_0_96", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DpVbta6xaUw_00001_647_1125.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DpVbta6xaUw_00001_647_1125.json new file mode 100644 index 0000000000000000000000000000000000000000..24d6576d5f6125a2c09d1a43894a137e8caef1f0 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DpVbta6xaUw_00001_647_1125.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_DpVbta6xaUw_00001_647_1125", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/DpVbta6xaUw_00001_647_1125", + "duration": 20.02 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DpVbta6xaUw_00001_647_1125_joint0_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DpVbta6xaUw_00001_647_1125_joint1_skel.jpg", + "time": 8.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DpVbta6xaUw_00001_647_1125_joint2_skel.jpg", + "time": 13.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/DpVbta6xaUw_00001_647_1125", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DsU0Cm6PmzE_00000_0_104.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DsU0Cm6PmzE_00000_0_104.json new file mode 100644 index 0000000000000000000000000000000000000000..c81d0d6ff54c395173a44812f378db8e7918802a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_DsU0Cm6PmzE_00000_0_104.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_DsU0Cm6PmzE_00000_0_104", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/DsU0Cm6PmzE_00000_0_104", + "duration": 4.338 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DsU0Cm6PmzE_00000_0_104_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DsU0Cm6PmzE_00000_0_104_joint1_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/DsU0Cm6PmzE_00000_0_104_joint2_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/DsU0Cm6PmzE_00000_0_104", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EAu_JbniRVw_00041_0_206.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EAu_JbniRVw_00041_0_206.json new file mode 100644 index 0000000000000000000000000000000000000000..b1c55497429206634035df72fa03d4546dcb2ffd --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EAu_JbniRVw_00041_0_206.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_EAu_JbniRVw_00041_0_206", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/EAu_JbniRVw_00041_0_206", + "duration": 8.36 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EAu_JbniRVw_00041_0_206_joint0_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/EAu_JbniRVw_00041_0_206", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EIibo7aTpys_00012_0_97.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EIibo7aTpys_00012_0_97.json new file mode 100644 index 0000000000000000000000000000000000000000..c07f915ab83cc2f77a3d35fb1a7332ede63f827a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EIibo7aTpys_00012_0_97.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_EIibo7aTpys_00012_0_97", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/EIibo7aTpys_00012_0_97", + "duration": 4.088 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EIibo7aTpys_00012_0_97_joint0_skel.jpg", + "time": 0.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/EIibo7aTpys_00012_0_97", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EJuVEQtoi7E_00013_0_77.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EJuVEQtoi7E_00013_0_77.json new file mode 100644 index 0000000000000000000000000000000000000000..97ee76757945317ae934ae3ad10097dedbf021af --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EJuVEQtoi7E_00013_0_77.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_EJuVEQtoi7E_00013_0_77", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/EJuVEQtoi7E_00013_0_77", + "duration": 2.67 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EJuVEQtoi7E_00013_0_77_joint0_skel.jpg", + "time": 0.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EJuVEQtoi7E_00013_0_77_joint1_skel.jpg", + "time": 1.2 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EJuVEQtoi7E_00013_0_77_joint2_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/EJuVEQtoi7E_00013_0_77", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ELV9T7pISm0_00001_0_145.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ELV9T7pISm0_00001_0_145.json new file mode 100644 index 0000000000000000000000000000000000000000..01a28b557a05e250b4483ef724b672f929c2a461 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ELV9T7pISm0_00001_0_145.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_ELV9T7pISm0_00001_0_145", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/ELV9T7pISm0_00001_0_145", + "duration": 4.872 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ELV9T7pISm0_00001_0_145_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ELV9T7pISm0_00001_0_145_joint1_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ELV9T7pISm0_00001_0_145_joint2_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/ELV9T7pISm0_00001_0_145", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EUA3T4A_5yA_00000_0_48.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EUA3T4A_5yA_00000_0_48.json new file mode 100644 index 0000000000000000000000000000000000000000..8bb1b40965576e039c535ac55b6d6dda43216deb --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EUA3T4A_5yA_00000_0_48.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_EUA3T4A_5yA_00000_0_48", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/EUA3T4A_5yA_00000_0_48", + "duration": 2.042 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EUA3T4A_5yA_00000_0_48_joint0_skel.jpg", + "time": 0.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/EUA3T4A_5yA_00000_0_48", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EiYsUTfJAw0_00058_0_189.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EiYsUTfJAw0_00058_0_189.json new file mode 100644 index 0000000000000000000000000000000000000000..d2c741c7853e3aedb7ce47b87bc77412133addcc --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EiYsUTfJAw0_00058_0_189.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_EiYsUTfJAw0_00058_0_189", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/EiYsUTfJAw0_00058_0_189", + "duration": 3.8 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EiYsUTfJAw0_00058_0_189_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EiYsUTfJAw0_00058_0_189_joint1_skel.jpg", + "time": 2.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/EiYsUTfJAw0_00058_0_189", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Eif-QsyWL7o_00069_0_234.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Eif-QsyWL7o_00069_0_234.json new file mode 100644 index 0000000000000000000000000000000000000000..e3806c2f96300a977bb137263a6e91ea39684ec0 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Eif-QsyWL7o_00069_0_234.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_Eif-QsyWL7o_00069_0_234", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/Eif-QsyWL7o_00069_0_234", + "duration": 4.7 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/Eif-QsyWL7o_00069_0_234_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/Eif-QsyWL7o_00069_0_234_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/Eif-QsyWL7o_00069_0_234", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EkMEGFBoYPc_00001_0_93.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EkMEGFBoYPc_00001_0_93.json new file mode 100644 index 0000000000000000000000000000000000000000..8bc9f6d2dfdde326874095451d5361a603e81398 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_EkMEGFBoYPc_00001_0_93.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_EkMEGFBoYPc_00001_0_93", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/EkMEGFBoYPc_00001_0_93", + "duration": 3.137 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/EkMEGFBoYPc_00001_0_93_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/EkMEGFBoYPc_00001_0_93", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ElIQTOQgWNw_00022_0_113.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ElIQTOQgWNw_00022_0_113.json new file mode 100644 index 0000000000000000000000000000000000000000..70564522c751c215c38290bcf98664669c304015 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ElIQTOQgWNw_00022_0_113.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_ElIQTOQgWNw_00022_0_113", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/ElIQTOQgWNw_00022_0_113", + "duration": 3.804 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ElIQTOQgWNw_00022_0_113_joint0_skel.jpg", + "time": 0.7 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ElIQTOQgWNw_00022_0_113_joint1_skel.jpg", + "time": 1.9 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/ElIQTOQgWNw_00022_0_113", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_F-bdFiGotmg_00012_0_123.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_F-bdFiGotmg_00012_0_123.json new file mode 100644 index 0000000000000000000000000000000000000000..a24fe4792c8d2ce185de066435cb3607eb9ca6e7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_F-bdFiGotmg_00012_0_123.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_F-bdFiGotmg_00012_0_123", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/F-bdFiGotmg_00012_0_123", + "duration": 4.171 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/F-bdFiGotmg_00012_0_123_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/F-bdFiGotmg_00012_0_123", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_F6M_9Gb3L8g_00006_173_472.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_F6M_9Gb3L8g_00006_173_472.json new file mode 100644 index 0000000000000000000000000000000000000000..3c6db4d51b4d8e77a696b23756c934d170a964eb --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_F6M_9Gb3L8g_00006_173_472.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_F6M_9Gb3L8g_00006_173_472", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/F6M_9Gb3L8g_00006_173_472", + "duration": 10.0 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/F6M_9Gb3L8g_00006_173_472_joint0_skel.jpg", + "time": 7.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/F6M_9Gb3L8g_00006_173_472", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FLnjU2oyn_0_00022_0_105.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FLnjU2oyn_0_00022_0_105.json new file mode 100644 index 0000000000000000000000000000000000000000..17ced5b2d908f4ac5b56bf814e0ad717931d1dcf --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FLnjU2oyn_0_00022_0_105.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_FLnjU2oyn_0_00022_0_105", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/FLnjU2oyn_0_00022_0_105", + "duration": 3.571 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/FLnjU2oyn_0_00022_0_105_joint0_skel.jpg", + "time": 0.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/FLnjU2oyn_0_00022_0_105", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FSxSuRaBLwQ_00063_0_225.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FSxSuRaBLwQ_00063_0_225.json new file mode 100644 index 0000000000000000000000000000000000000000..4002858b85b43d511d8d8757da2f32c18a56d301 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FSxSuRaBLwQ_00063_0_225.json @@ -0,0 +1,44 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_FSxSuRaBLwQ_00063_0_225", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/FSxSuRaBLwQ_00063_0_225", + "duration": 9.427 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/FSxSuRaBLwQ_00063_0_225_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/FSxSuRaBLwQ_00063_0_225_joint1_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/FSxSuRaBLwQ_00063_0_225_joint2_skel.jpg", + "time": 5.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/FSxSuRaBLwQ_00063_0_225_joint3_skel.jpg", + "time": 7.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/FSxSuRaBLwQ_00063_0_225", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FcFGSUx2FPM_00009_0_72.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FcFGSUx2FPM_00009_0_72.json new file mode 100644 index 0000000000000000000000000000000000000000..e829a89524265fa41fcea605edf62807c0f4dea9 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_FcFGSUx2FPM_00009_0_72.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_FcFGSUx2FPM_00009_0_72", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/FcFGSUx2FPM_00009_0_72", + "duration": 2.92 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/FcFGSUx2FPM_00009_0_72_joint0_skel.jpg", + "time": 1.3 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/FcFGSUx2FPM_00009_0_72", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HMYRSj3Rju4_00100_0_123.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HMYRSj3Rju4_00100_0_123.json new file mode 100644 index 0000000000000000000000000000000000000000..3af4242fe361779007b563daeca53d3ef0cd3712 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HMYRSj3Rju4_00100_0_123.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_HMYRSj3Rju4_00100_0_123", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/HMYRSj3Rju4_00100_0_123", + "duration": 4.144 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/HMYRSj3Rju4_00100_0_123_joint0_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/HMYRSj3Rju4_00100_0_123", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HXBUzJigzrg_00028_0_960.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HXBUzJigzrg_00028_0_960.json new file mode 100644 index 0000000000000000000000000000000000000000..7455f55be926a91cb442dc5060094aace3d8d026 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HXBUzJigzrg_00028_0_960.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_HXBUzJigzrg_00028_0_960", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/HXBUzJigzrg_00028_0_960", + "duration": 16.306 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/HXBUzJigzrg_00028_0_960_joint0_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/HXBUzJigzrg_00028_0_960_joint1_skel.jpg", + "time": 9.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/HXBUzJigzrg_00028_0_960", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HbGLl85N7Cw_00001_0_65.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HbGLl85N7Cw_00001_0_65.json new file mode 100644 index 0000000000000000000000000000000000000000..8852bd65887f327bfbce5ba2e88ac07346dfd811 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HbGLl85N7Cw_00001_0_65.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_HbGLl85N7Cw_00001_0_65", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/HbGLl85N7Cw_00001_0_65", + "duration": 2.64 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/HbGLl85N7Cw_00001_0_65_joint0_skel.jpg", + "time": 0.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/HbGLl85N7Cw_00001_0_65", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HfM2BjLPm1w_00023_0_175.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HfM2BjLPm1w_00023_0_175.json new file mode 100644 index 0000000000000000000000000000000000000000..92cfc018c579e0d07d49fa3784b5bf1f7cd00005 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_HfM2BjLPm1w_00023_0_175.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_HfM2BjLPm1w_00023_0_175", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/HfM2BjLPm1w_00023_0_175", + "duration": 5.873 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/HfM2BjLPm1w_00023_0_175_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/HfM2BjLPm1w_00023_0_175", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IHSaR35NJuM_00001_1062_1406.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IHSaR35NJuM_00001_1062_1406.json new file mode 100644 index 0000000000000000000000000000000000000000..7f2cec7c2beb22af248926519d254f200eb8b453 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IHSaR35NJuM_00001_1062_1406.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_IHSaR35NJuM_00001_1062_1406", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/IHSaR35NJuM_00001_1062_1406", + "duration": 11.545 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IHSaR35NJuM_00001_1062_1406_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IHSaR35NJuM_00001_1062_1406_joint1_skel.jpg", + "time": 6.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IHSaR35NJuM_00001_1062_1406_joint2_skel.jpg", + "time": 9.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/IHSaR35NJuM_00001_1062_1406", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IRJ014GJbYw_00073_0_102.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IRJ014GJbYw_00073_0_102.json new file mode 100644 index 0000000000000000000000000000000000000000..35c6ed37f1bfa5e661ef88f5d421437eedea9682 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IRJ014GJbYw_00073_0_102.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_IRJ014GJbYw_00073_0_102", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/IRJ014GJbYw_00073_0_102", + "duration": 4.338 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is standing at a kitchen counter, reaching for an object, and turning to unscrew a bottle.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IRJ014GJbYw_00073_0_102_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/IRJ014GJbYw_00073_0_102", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IXdgnxngfs4_00007_16_98.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IXdgnxngfs4_00007_16_98.json new file mode 100644 index 0000000000000000000000000000000000000000..9e711cd0d85042e92f0f6cc38242608d95e4e32a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IXdgnxngfs4_00007_16_98.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_IXdgnxngfs4_00007_16_98", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/IXdgnxngfs4_00007_16_98", + "duration": 3.32 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IXdgnxngfs4_00007_16_98_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/IXdgnxngfs4_00007_16_98", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IbuZuIWQ16k_00016_0_823.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IbuZuIWQ16k_00016_0_823.json new file mode 100644 index 0000000000000000000000000000000000000000..a19e812e1608105c5fa0c9deb653bb913e2909bf --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IbuZuIWQ16k_00016_0_823.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_IbuZuIWQ16k_00016_0_823", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/IbuZuIWQ16k_00016_0_823", + "duration": 13.764 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IbuZuIWQ16k_00016_0_823_joint0_skel.jpg", + "time": 5.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IbuZuIWQ16k_00016_0_823_joint1_skel.jpg", + "time": 10.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/IbuZuIWQ16k_00016_0_823", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Ir83IWRVTOw_00004_0_242.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Ir83IWRVTOw_00004_0_242.json new file mode 100644 index 0000000000000000000000000000000000000000..0d0db140110214c086308903059f259079337ade --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Ir83IWRVTOw_00004_0_242.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_Ir83IWRVTOw_00004_0_242", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/Ir83IWRVTOw_00004_0_242", + "duration": 10.084 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/Ir83IWRVTOw_00004_0_242_joint0_skel.jpg", + "time": 7.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/Ir83IWRVTOw_00004_0_242", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IvUYGasf4-g_00053_0_113.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IvUYGasf4-g_00053_0_113.json new file mode 100644 index 0000000000000000000000000000000000000000..b64a72c29ad9c067bf3048741e70796068c69d1d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_IvUYGasf4-g_00053_0_113.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_IvUYGasf4-g_00053_0_113", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/IvUYGasf4-g_00053_0_113", + "duration": 3.904 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IvUYGasf4-g_00053_0_113_joint0_skel.jpg", + "time": 0.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IvUYGasf4-g_00053_0_113_joint1_skel.jpg", + "time": 1.4 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/IvUYGasf4-g_00053_0_113_joint2_skel.jpg", + "time": 2.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/IvUYGasf4-g_00053_0_113", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Iy5prfkaOAY_00001_238_354.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Iy5prfkaOAY_00001_238_354.json new file mode 100644 index 0000000000000000000000000000000000000000..e88a5c814d7e90fb855cf47decf81be833f310d2 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_Iy5prfkaOAY_00001_238_354.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_Iy5prfkaOAY_00001_238_354", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/Iy5prfkaOAY_00001_238_354", + "duration": 3.938 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/Iy5prfkaOAY_00001_238_354_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/Iy5prfkaOAY_00001_238_354", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J27DCHIJauw_00027_0_82.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J27DCHIJauw_00027_0_82.json new file mode 100644 index 0000000000000000000000000000000000000000..23243ecd673c4d037b0630e27363f9d495d732fd --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J27DCHIJauw_00027_0_82.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_J27DCHIJauw_00027_0_82", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/J27DCHIJauw_00027_0_82", + "duration": 2.77 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/J27DCHIJauw_00027_0_82_joint0_skel.jpg", + "time": 0.8 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/J27DCHIJauw_00027_0_82_joint1_skel.jpg", + "time": 1.6 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/J27DCHIJauw_00027_0_82", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J5p4Zmu7zDQ_00023_84_361.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J5p4Zmu7zDQ_00023_84_361.json new file mode 100644 index 0000000000000000000000000000000000000000..06cef846578e782b7c4601c27d818f2be1a5e4c0 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J5p4Zmu7zDQ_00023_84_361.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_J5p4Zmu7zDQ_00023_84_361", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/J5p4Zmu7zDQ_00023_84_361", + "duration": 9.377 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/J5p4Zmu7zDQ_00023_84_361_joint0_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/J5p4Zmu7zDQ_00023_84_361", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JY1mxnXCeF8_00003_511_598.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JY1mxnXCeF8_00003_511_598.json new file mode 100644 index 0000000000000000000000000000000000000000..08eff5e3ec0c710d19e1ca060c0abaa9598ecb84 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JY1mxnXCeF8_00003_511_598.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_JY1mxnXCeF8_00003_511_598", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/JY1mxnXCeF8_00003_511_598", + "duration": 2.904 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/JY1mxnXCeF8_00003_511_598_joint0_skel.jpg", + "time": 0.7 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/JY1mxnXCeF8_00003_511_598_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/JY1mxnXCeF8_00003_511_598", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JYL3s66-0W4_00056_288_374.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JYL3s66-0W4_00056_288_374.json new file mode 100644 index 0000000000000000000000000000000000000000..317e3e54e6ac19a0da1a4eed4a90a127272654de --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JYL3s66-0W4_00056_288_374.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_JYL3s66-0W4_00056_288_374", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/JYL3s66-0W4_00056_288_374", + "duration": 3.56 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/JYL3s66-0W4_00056_288_374_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/JYL3s66-0W4_00056_288_374_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/JYL3s66-0W4_00056_288_374", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J_4NWbKHP0I_00012_0_82.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J_4NWbKHP0I_00012_0_82.json new file mode 100644 index 0000000000000000000000000000000000000000..4d921f44810c2e07e76408c06108aecc8b52e35a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_J_4NWbKHP0I_00012_0_82.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_J_4NWbKHP0I_00012_0_82", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/J_4NWbKHP0I_00012_0_82", + "duration": 2.77 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/J_4NWbKHP0I_00012_0_82_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/J_4NWbKHP0I_00012_0_82_joint1_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/J_4NWbKHP0I_00012_0_82", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JvV640xrJ28_00026_0_84.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JvV640xrJ28_00026_0_84.json new file mode 100644 index 0000000000000000000000000000000000000000..d64e1023143330c0e8a92c804e140c37204c0407 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_JvV640xrJ28_00026_0_84.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_JvV640xrJ28_00026_0_84", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/JvV640xrJ28_00026_0_84", + "duration": 3.503 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/JvV640xrJ28_00026_0_84_joint0_skel.jpg", + "time": 0.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/JvV640xrJ28_00026_0_84", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_K8hCBLTuVyk_00045_0_137.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_K8hCBLTuVyk_00045_0_137.json new file mode 100644 index 0000000000000000000000000000000000000000..ce5ec9448a67fe7a6fa1ad7dfe6727c378e65eda --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_K8hCBLTuVyk_00045_0_137.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_K8hCBLTuVyk_00045_0_137", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/K8hCBLTuVyk_00045_0_137", + "duration": 4.638 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/K8hCBLTuVyk_00045_0_137_joint0_skel.jpg", + "time": 3.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/K8hCBLTuVyk_00045_0_137", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRY1aqBGTCE_00057_0_57.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRY1aqBGTCE_00057_0_57.json new file mode 100644 index 0000000000000000000000000000000000000000..205cb35446b1d0e5bb9852214407e74ada947132 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRY1aqBGTCE_00057_0_57.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_KRY1aqBGTCE_00057_0_57", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/KRY1aqBGTCE_00057_0_57", + "duration": 2.375 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/KRY1aqBGTCE_00057_0_57_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/KRY1aqBGTCE_00057_0_57", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRheAHqm4SU_00001_423_720.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRheAHqm4SU_00001_423_720.json new file mode 100644 index 0000000000000000000000000000000000000000..83eb4958e266111ece20e32fa4eef2bab04a56f6 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRheAHqm4SU_00001_423_720.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_KRheAHqm4SU_00001_423_720", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/KRheAHqm4SU_00001_423_720", + "duration": 9.918 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/KRheAHqm4SU_00001_423_720_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/KRheAHqm4SU_00001_423_720_joint1_skel.jpg", + "time": 6.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/KRheAHqm4SU_00001_423_720", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRheAHqm4SU_00006_477_633.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRheAHqm4SU_00006_477_633.json new file mode 100644 index 0000000000000000000000000000000000000000..f031df1329535d1c7436246a0e915cb3b1776b77 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KRheAHqm4SU_00006_477_633.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_KRheAHqm4SU_00006_477_633", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/KRheAHqm4SU_00006_477_633", + "duration": 5.206 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/KRheAHqm4SU_00006_477_633_joint0_skel.jpg", + "time": 3.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/KRheAHqm4SU_00006_477_633", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KU5e1HH-j-c_00018_0_135.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KU5e1HH-j-c_00018_0_135.json new file mode 100644 index 0000000000000000000000000000000000000000..33cd82accc1d68fe18a5d9d498c6872bf215b908 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_KU5e1HH-j-c_00018_0_135.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_KU5e1HH-j-c_00018_0_135", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/KU5e1HH-j-c_00018_0_135", + "duration": 4.634 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/KU5e1HH-j-c_00018_0_135_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/KU5e1HH-j-c_00018_0_135", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L1rESfCgUZo_00010_0_117.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L1rESfCgUZo_00010_0_117.json new file mode 100644 index 0000000000000000000000000000000000000000..5b02a96a99fa88df8d19ce6b9a1b730410f71245 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L1rESfCgUZo_00010_0_117.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_L1rESfCgUZo_00010_0_117", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/L1rESfCgUZo_00010_0_117", + "duration": 3.938 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/L1rESfCgUZo_00010_0_117_joint0_skel.jpg", + "time": 0.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/L1rESfCgUZo_00010_0_117_joint1_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/L1rESfCgUZo_00010_0_117", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L44lwaZ-qAM_00004_0_170.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L44lwaZ-qAM_00004_0_170.json new file mode 100644 index 0000000000000000000000000000000000000000..55312329123e63dd9f2d4d3a6396d6724ae5b278 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L44lwaZ-qAM_00004_0_170.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_L44lwaZ-qAM_00004_0_170", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/L44lwaZ-qAM_00004_0_170", + "duration": 7.435 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/L44lwaZ-qAM_00004_0_170_joint0_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/L44lwaZ-qAM_00004_0_170", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L72G8TLtTCk_00014_0_105.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L72G8TLtTCk_00014_0_105.json new file mode 100644 index 0000000000000000000000000000000000000000..29a8353bcf4a5cba0f9d59dcfec56c1e288e5241 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_L72G8TLtTCk_00014_0_105.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_L72G8TLtTCk_00014_0_105", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/L72G8TLtTCk_00014_0_105", + "duration": 4.421 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/L72G8TLtTCk_00014_0_105_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/L72G8TLtTCk_00014_0_105", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LALTJJyO2js_00000_0_66.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LALTJJyO2js_00000_0_66.json new file mode 100644 index 0000000000000000000000000000000000000000..29eb564d6cb35c784c989b176bd701ff866ca8e4 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LALTJJyO2js_00000_0_66.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LALTJJyO2js_00000_0_66", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LALTJJyO2js_00000_0_66", + "duration": 2.68 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is walking through a dimly lit hallway lined with lockers and stopping to place a bag inside one of them.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LALTJJyO2js_00000_0_66_joint0_skel.jpg", + "time": 0.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LALTJJyO2js_00000_0_66_joint1_skel.jpg", + "time": 1.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LALTJJyO2js_00000_0_66", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LDM9aXVAoOw_00002_0_694.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LDM9aXVAoOw_00002_0_694.json new file mode 100644 index 0000000000000000000000000000000000000000..c5e10052635ec4d833974bb4863f01af4e469521 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LDM9aXVAoOw_00002_0_694.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LDM9aXVAoOw_00002_0_694", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LDM9aXVAoOw_00002_0_694", + "duration": 13.9 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LDM9aXVAoOw_00002_0_694_joint0_skel.jpg", + "time": 8.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LDM9aXVAoOw_00002_0_694_joint1_skel.jpg", + "time": 10.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LDM9aXVAoOw_00002_0_694_joint2_skel.jpg", + "time": 12.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LDM9aXVAoOw_00002_0_694", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LEppqung-TQ_00006_0_109.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LEppqung-TQ_00006_0_109.json new file mode 100644 index 0000000000000000000000000000000000000000..94fb7e82a7b15ff0cebd7f78fd4a63fd5f66e685 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LEppqung-TQ_00006_0_109.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LEppqung-TQ_00006_0_109", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LEppqung-TQ_00006_0_109", + "duration": 4.4 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LEppqung-TQ_00006_0_109_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LEppqung-TQ_00006_0_109", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LGoX7vNg3IM_00026_592_719.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LGoX7vNg3IM_00026_592_719.json new file mode 100644 index 0000000000000000000000000000000000000000..aeda5b5ae838aad629e4db1c83186327014616cd --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LGoX7vNg3IM_00026_592_719.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LGoX7vNg3IM_00026_592_719", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LGoX7vNg3IM_00026_592_719", + "duration": 2.156 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LGoX7vNg3IM_00026_592_719_joint0_skel.jpg", + "time": 0.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LGoX7vNg3IM_00026_592_719", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LNvgFIUv9hI_00023_0_59.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LNvgFIUv9hI_00023_0_59.json new file mode 100644 index 0000000000000000000000000000000000000000..72babe7652b1b1670954270388b9162eb9013f1b --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LNvgFIUv9hI_00023_0_59.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LNvgFIUv9hI_00023_0_59", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LNvgFIUv9hI_00023_0_59", + "duration": 2.44 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LNvgFIUv9hI_00023_0_59_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LNvgFIUv9hI_00023_0_59_joint1_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LNvgFIUv9hI_00023_0_59", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LOWyPvH0210_00002_0_103.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LOWyPvH0210_00002_0_103.json new file mode 100644 index 0000000000000000000000000000000000000000..1986047238549026a4eab593cd5cd24730dca812 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LOWyPvH0210_00002_0_103.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LOWyPvH0210_00002_0_103", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LOWyPvH0210_00002_0_103", + "duration": 3.537 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LOWyPvH0210_00002_0_103_joint0_skel.jpg", + "time": 0.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LOWyPvH0210_00002_0_103_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LOWyPvH0210_00002_0_103", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LTozxHr17ME_00024_0_119.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LTozxHr17ME_00024_0_119.json new file mode 100644 index 0000000000000000000000000000000000000000..7869d875b5b2a9310a53d36ec6f484560d052cf6 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LTozxHr17ME_00024_0_119.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LTozxHr17ME_00024_0_119", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LTozxHr17ME_00024_0_119", + "duration": 4.004 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is transitioning from a focused run across the baseball field to a stationary observation of the game.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LTozxHr17ME_00024_0_119_joint0_skel.jpg", + "time": 1.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LTozxHr17ME_00024_0_119", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LX5G4VkRH3o_00035_0_74.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LX5G4VkRH3o_00035_0_74.json new file mode 100644 index 0000000000000000000000000000000000000000..74a594903f2bcf98d5630d82e7d70a5606522787 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LX5G4VkRH3o_00035_0_74.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LX5G4VkRH3o_00035_0_74", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LX5G4VkRH3o_00035_0_74", + "duration": 3.167 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LX5G4VkRH3o_00035_0_74_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LX5G4VkRH3o_00035_0_74_joint1_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LX5G4VkRH3o_00035_0_74_joint2_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LX5G4VkRH3o_00035_0_74", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LZ9kV0wNDrY_00011_393_503.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LZ9kV0wNDrY_00011_393_503.json new file mode 100644 index 0000000000000000000000000000000000000000..8aa428fb8b574c9fc9631b6cf3c4f4ccdd350f1f --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LZ9kV0wNDrY_00011_393_503.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LZ9kV0wNDrY_00011_393_503", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LZ9kV0wNDrY_00011_393_503", + "duration": 3.641 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LZ9kV0wNDrY_00011_393_503_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LZ9kV0wNDrY_00011_393_503", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LtVkTfmcyOw_00048_0_99.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LtVkTfmcyOw_00048_0_99.json new file mode 100644 index 0000000000000000000000000000000000000000..ca52ef3e1242799de092d56fae3619c0dafbbc5c --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_LtVkTfmcyOw_00048_0_99.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_LtVkTfmcyOw_00048_0_99", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/LtVkTfmcyOw_00048_0_99", + "duration": 3.337 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "My Thought Process on Summarizing the Text\n\nOkay, let's break this down systematically. First, I need to understand the task precisely: summarize, in a single sentence, starting with \"The person is...\", and avoiding any body part references. It seems there's a slight contradiction, the prompt implies the output will be the full paragraph but the task is to summarize. I'll assume that means \"give me the full summary sentence,\" not the original text, as that is the purpose of summarization. The input describes a child's actions: standing still, walking, turning, and moving. I need to capture the essence of this sequence. My initial thought is \"The person is a child who stands still in a hallway before walking forward and turning left toward the dining table.\" Let's check against my constraints: single sentence? Check. Starts correctly? Check. No body parts referenced? Check. Accuracy? Yes. However, I can refine it further to improve flow and sophistication. I could replace \"before\" with a more formal transition word like \"transitions.\" So, I now revise: \"The person is a child who transitions from standing still in a hallway to walking forward and turning left toward the dining table.\" I've verified that this meets all the criteria: One sentence, starting correctly, avoids body parts, and I think it accurately captures the narrative flow. It's concise and comprehensive. Now, I will construct the final output.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/LtVkTfmcyOw_00048_0_99_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/LtVkTfmcyOw_00048_0_99", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_M3oVDmDIHek_00095_0_134.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_M3oVDmDIHek_00095_0_134.json new file mode 100644 index 0000000000000000000000000000000000000000..fc3a7faad5626dab090c6a80a5683498bec582b7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_M3oVDmDIHek_00095_0_134.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_M3oVDmDIHek_00095_0_134", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/M3oVDmDIHek_00095_0_134", + "duration": 2.701 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/M3oVDmDIHek_00095_0_134_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/M3oVDmDIHek_00095_0_134_joint1_skel.jpg", + "time": 1.6 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/M3oVDmDIHek_00095_0_134", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__1tzdCdC4UY_00005_77_203.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__1tzdCdC4UY_00005_77_203.json new file mode 100644 index 0000000000000000000000000000000000000000..60fc388e926cfdb774fbcb4c89ae9f7b5d395d94 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__1tzdCdC4UY_00005_77_203.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL__1tzdCdC4UY_00005_77_203", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/_1tzdCdC4UY_00005_77_203", + "duration": 5.259 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is addressing the audience while standing, then leans down to retrieve a piece of fabric from a table and begins to unfold it.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_1tzdCdC4UY_00005_77_203_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_1tzdCdC4UY_00005_77_203_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/_1tzdCdC4UY_00005_77_203", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__MXhQXP5hEI_00002_0_662.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__MXhQXP5hEI_00002_0_662.json new file mode 100644 index 0000000000000000000000000000000000000000..8dc0bc32b292ef54a50e7cd16a14d018b065619b --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__MXhQXP5hEI_00002_0_662.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL__MXhQXP5hEI_00002_0_662", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/_MXhQXP5hEI_00002_0_662", + "duration": 11.08 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_MXhQXP5hEI_00002_0_662_joint0_skel.jpg", + "time": 3.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_MXhQXP5hEI_00002_0_662_joint1_skel.jpg", + "time": 6.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/_MXhQXP5hEI_00002_0_662", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__dFAYYMJbWk_00193_268_382.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__dFAYYMJbWk_00193_268_382.json new file mode 100644 index 0000000000000000000000000000000000000000..7ede2b0159ccb5fcaf1ee176c7019bb0ed578bff --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__dFAYYMJbWk_00193_268_382.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL__dFAYYMJbWk_00193_268_382", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/_dFAYYMJbWk_00193_268_382", + "duration": 4.6 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_dFAYYMJbWk_00193_268_382_joint0_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/_dFAYYMJbWk_00193_268_382", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__eyx3-8oIr0_00009_0_234.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__eyx3-8oIr0_00009_0_234.json new file mode 100644 index 0000000000000000000000000000000000000000..36d4ef777a8e2c928e955a75753676dd24d8de63 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__eyx3-8oIr0_00009_0_234.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL__eyx3-8oIr0_00009_0_234", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/_eyx3-8oIr0_00009_0_234", + "duration": 7.842 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_eyx3-8oIr0_00009_0_234_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_eyx3-8oIr0_00009_0_234_joint1_skel.jpg", + "time": 5.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/_eyx3-8oIr0_00009_0_234", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__fIznC6ZIX0_00032_0_67.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__fIznC6ZIX0_00032_0_67.json new file mode 100644 index 0000000000000000000000000000000000000000..b46b173a263d54cfd741d3246d4e2fd636e3df25 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__fIznC6ZIX0_00032_0_67.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL__fIznC6ZIX0_00032_0_67", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/_fIznC6ZIX0_00032_0_67", + "duration": 2.267 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_fIznC6ZIX0_00032_0_67_joint0_skel.jpg", + "time": 0.8 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_fIznC6ZIX0_00032_0_67_joint1_skel.jpg", + "time": 1.4 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/_fIznC6ZIX0_00032_0_67", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__oYIT7I3avY_00018_947_1296.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__oYIT7I3avY_00018_947_1296.json new file mode 100644 index 0000000000000000000000000000000000000000..eb2762362942ea9e4f45cb5c653c27ae13fb0366 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL__oYIT7I3avY_00018_947_1296.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL__oYIT7I3avY_00018_947_1296", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/_oYIT7I3avY_00018_947_1296", + "duration": 11.679 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_oYIT7I3avY_00018_947_1296_joint0_skel.jpg", + "time": 3.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/_oYIT7I3avY_00018_947_1296_joint1_skel.jpg", + "time": 6.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/_oYIT7I3avY_00018_947_1296", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_a500SZyQFxQ_00014_0_65.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_a500SZyQFxQ_00014_0_65.json new file mode 100644 index 0000000000000000000000000000000000000000..876701e8fd854afe523977577599a83115da8fd7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_a500SZyQFxQ_00014_0_65.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_a500SZyQFxQ_00014_0_65", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/a500SZyQFxQ_00014_0_65", + "duration": 2.711 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/a500SZyQFxQ_00014_0_65_joint0_skel.jpg", + "time": 0.6 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/a500SZyQFxQ_00014_0_65", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_aDuB1arf8aE_00030_0_111.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_aDuB1arf8aE_00030_0_111.json new file mode 100644 index 0000000000000000000000000000000000000000..1e400087261e470e1e8eb93480c0cca127a1f01d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_aDuB1arf8aE_00030_0_111.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_aDuB1arf8aE_00030_0_111", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/aDuB1arf8aE_00030_0_111", + "duration": 3.771 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/aDuB1arf8aE_00030_0_111_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/aDuB1arf8aE_00030_0_111_joint1_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/aDuB1arf8aE_00030_0_111_joint2_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/aDuB1arf8aE_00030_0_111", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_aEFB3NGhx9E_00034_0_146.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_aEFB3NGhx9E_00034_0_146.json new file mode 100644 index 0000000000000000000000000000000000000000..a4ba569e6756c41d1663a6dfa45ec36382d401cf --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_aEFB3NGhx9E_00034_0_146.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_aEFB3NGhx9E_00034_0_146", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/aEFB3NGhx9E_00034_0_146", + "duration": 4.905 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/aEFB3NGhx9E_00034_0_146_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/aEFB3NGhx9E_00034_0_146_joint1_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/aEFB3NGhx9E_00034_0_146", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ae4jA0iRvtg_00030_0_120.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ae4jA0iRvtg_00030_0_120.json new file mode 100644 index 0000000000000000000000000000000000000000..d60e5ad1df2780edd013eb5cb2855ffa1303ab4c --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ae4jA0iRvtg_00030_0_120.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_ae4jA0iRvtg_00030_0_120", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/ae4jA0iRvtg_00030_0_120", + "duration": 2.42 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ae4jA0iRvtg_00030_0_120_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ae4jA0iRvtg_00030_0_120_joint1_skel.jpg", + "time": 1.2 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ae4jA0iRvtg_00030_0_120_joint2_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/ae4jA0iRvtg_00030_0_120", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_b9-gJuDgKnU_00012_0_86.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_b9-gJuDgKnU_00012_0_86.json new file mode 100644 index 0000000000000000000000000000000000000000..0692a72bbdd1ecf9cd17d983937f68d6e861c16a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_b9-gJuDgKnU_00012_0_86.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_b9-gJuDgKnU_00012_0_86", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/b9-gJuDgKnU_00012_0_86", + "duration": 3.48 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/b9-gJuDgKnU_00012_0_86_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/b9-gJuDgKnU_00012_0_86_joint1_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/b9-gJuDgKnU_00012_0_86", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bFX5Jd2bHS8_00001_0_135.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bFX5Jd2bHS8_00001_0_135.json new file mode 100644 index 0000000000000000000000000000000000000000..4adad6262c07396c4b0e3ce76686d32037eecc7c --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bFX5Jd2bHS8_00001_0_135.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_bFX5Jd2bHS8_00001_0_135", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/bFX5Jd2bHS8_00001_0_135", + "duration": 4.538 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bFX5Jd2bHS8_00001_0_135_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bFX5Jd2bHS8_00001_0_135_joint1_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/bFX5Jd2bHS8_00001_0_135", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bN-o8SRUdC4_00010_67_186.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bN-o8SRUdC4_00010_67_186.json new file mode 100644 index 0000000000000000000000000000000000000000..5f9cc35bc30f2829a4f47798dd0a615ba21405f7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bN-o8SRUdC4_00010_67_186.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_bN-o8SRUdC4_00010_67_186", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/bN-o8SRUdC4_00010_67_186", + "duration": 4.0 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bN-o8SRUdC4_00010_67_186_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bN-o8SRUdC4_00010_67_186_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/bN-o8SRUdC4_00010_67_186", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bQoRtsNwRCo_00026_711_840.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bQoRtsNwRCo_00026_711_840.json new file mode 100644 index 0000000000000000000000000000000000000000..fe5e3680c64b5e353e2e168862f89e7018d68175 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bQoRtsNwRCo_00026_711_840.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_bQoRtsNwRCo_00026_711_840", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/bQoRtsNwRCo_00026_711_840", + "duration": 4.372 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bQoRtsNwRCo_00026_711_840_joint0_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/bQoRtsNwRCo_00026_711_840", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bS5ve8rG-Ik_00052_0_93.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bS5ve8rG-Ik_00052_0_93.json new file mode 100644 index 0000000000000000000000000000000000000000..f75639b1215e47aa8f91f5ab66422a33c96cb477 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bS5ve8rG-Ik_00052_0_93.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_bS5ve8rG-Ik_00052_0_93", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/bS5ve8rG-Ik_00052_0_93", + "duration": 3.137 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bS5ve8rG-Ik_00052_0_93_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bS5ve8rG-Ik_00052_0_93_joint1_skel.jpg", + "time": 1.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/bS5ve8rG-Ik_00052_0_93", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bSdMZBG-tDg_00016_0_359.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bSdMZBG-tDg_00016_0_359.json new file mode 100644 index 0000000000000000000000000000000000000000..0b6c70293026279ebc9a606f22a8ef313f419cb4 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bSdMZBG-tDg_00016_0_359.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_bSdMZBG-tDg_00016_0_359", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/bSdMZBG-tDg_00016_0_359", + "duration": 6.011 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is transitioning from a stationary position at the garage to riding a skateboard smoothly down the driveway and onto the street while maintaining a forward-leaning orientation for momentum.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bSdMZBG-tDg_00016_0_359_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/bSdMZBG-tDg_00016_0_359", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bohbtpZJkXo_00042_38_96.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bohbtpZJkXo_00042_38_96.json new file mode 100644 index 0000000000000000000000000000000000000000..f6c7fcf9e7ffeec92133d0177e36385dec56131e --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bohbtpZJkXo_00042_38_96.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_bohbtpZJkXo_00042_38_96", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/bohbtpZJkXo_00042_38_96", + "duration": 2.36 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bohbtpZJkXo_00042_38_96_joint0_skel.jpg", + "time": 0.7 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/bohbtpZJkXo_00042_38_96", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bwr6x5xT9nE_00001_120_711.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bwr6x5xT9nE_00001_120_711.json new file mode 100644 index 0000000000000000000000000000000000000000..24de30d12b56da78472abc9ff60da5ab3aebdab2 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_bwr6x5xT9nE_00001_120_711.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_bwr6x5xT9nE_00001_120_711", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/bwr6x5xT9nE_00001_120_711", + "duration": 9.888 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bwr6x5xT9nE_00001_120_711_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/bwr6x5xT9nE_00001_120_711_joint1_skel.jpg", + "time": 6.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/bwr6x5xT9nE_00001_120_711", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_c03xSP9PzyA_00044_0_101.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_c03xSP9PzyA_00044_0_101.json new file mode 100644 index 0000000000000000000000000000000000000000..d2a9106f8e9f8c4aec7ae0d6f6be1b598b343567 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_c03xSP9PzyA_00044_0_101.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_c03xSP9PzyA_00044_0_101", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/c03xSP9PzyA_00044_0_101", + "duration": 4.08 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/c03xSP9PzyA_00044_0_101_joint0_skel.jpg", + "time": 1.2 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/c03xSP9PzyA_00044_0_101_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/c03xSP9PzyA_00044_0_101", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_c3Dh8n1qyCY_00009_0_66.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_c3Dh8n1qyCY_00009_0_66.json new file mode 100644 index 0000000000000000000000000000000000000000..24f940da280022efc59653e588a1c174a893830e --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_c3Dh8n1qyCY_00009_0_66.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_c3Dh8n1qyCY_00009_0_66", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/c3Dh8n1qyCY_00009_0_66", + "duration": 2.236 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/c3Dh8n1qyCY_00009_0_66_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/c3Dh8n1qyCY_00009_0_66", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cAEYHc5dzfs_00003_297_392.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cAEYHc5dzfs_00003_297_392.json new file mode 100644 index 0000000000000000000000000000000000000000..901485b73225770fa39821df4438a93931918955 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cAEYHc5dzfs_00003_297_392.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_cAEYHc5dzfs_00003_297_392", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/cAEYHc5dzfs_00003_297_392", + "duration": 3.204 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/cAEYHc5dzfs_00003_297_392_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/cAEYHc5dzfs_00003_297_392_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/cAEYHc5dzfs_00003_297_392", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cEdo6STidC0_00023_361_458.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cEdo6STidC0_00023_361_458.json new file mode 100644 index 0000000000000000000000000000000000000000..5e2e2049d57518cb7226bef8c7af86cd517ab6d6 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cEdo6STidC0_00023_361_458.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_cEdo6STidC0_00023_361_458", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/cEdo6STidC0_00023_361_458", + "duration": 3.27 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/cEdo6STidC0_00023_361_458_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/cEdo6STidC0_00023_361_458_joint1_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/cEdo6STidC0_00023_361_458", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cUUQtoBnAJU_00000_2062_2222.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cUUQtoBnAJU_00000_2062_2222.json new file mode 100644 index 0000000000000000000000000000000000000000..d3937559a4b74e6e1a58714b181017e8b7413a72 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_cUUQtoBnAJU_00000_2062_2222.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_cUUQtoBnAJU_00000_2062_2222", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/cUUQtoBnAJU_00000_2062_2222", + "duration": 2.683 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/cUUQtoBnAJU_00000_2062_2222_joint0_skel.jpg", + "time": 0.8 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/cUUQtoBnAJU_00000_2062_2222_joint1_skel.jpg", + "time": 1.6 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/cUUQtoBnAJU_00000_2062_2222", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_d3uu7C26flM_00001_0_83.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_d3uu7C26flM_00001_0_83.json new file mode 100644 index 0000000000000000000000000000000000000000..7382842cf8d1d8a1136a66c805c90c09a72f1781 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_d3uu7C26flM_00001_0_83.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_d3uu7C26flM_00001_0_83", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/d3uu7C26flM_00001_0_83", + "duration": 2.803 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/d3uu7C26flM_00001_0_83_joint0_skel.jpg", + "time": 0.8 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/d3uu7C26flM_00001_0_83_joint1_skel.jpg", + "time": 1.7 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/d3uu7C26flM_00001_0_83", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_htSmYNfT8QA_00045_0_209.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_htSmYNfT8QA_00045_0_209.json new file mode 100644 index 0000000000000000000000000000000000000000..8c3012a3be0e3501417730d646caa3075b95fabe --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_htSmYNfT8QA_00045_0_209.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_htSmYNfT8QA_00045_0_209", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/htSmYNfT8QA_00045_0_209", + "duration": 6.983 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/htSmYNfT8QA_00045_0_209_joint0_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/htSmYNfT8QA_00045_0_209_joint1_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/htSmYNfT8QA_00045_0_209", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_htSmYNfT8QA_00054_0_195.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_htSmYNfT8QA_00054_0_195.json new file mode 100644 index 0000000000000000000000000000000000000000..9c248b2c9e04211376acdc69b4a8d2aed6f7a318 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_htSmYNfT8QA_00054_0_195.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_htSmYNfT8QA_00054_0_195", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/htSmYNfT8QA_00054_0_195", + "duration": 6.541 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/htSmYNfT8QA_00054_0_195_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/htSmYNfT8QA_00054_0_195_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/htSmYNfT8QA_00054_0_195", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_httbW7TujP0_00057_0_170.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_httbW7TujP0_00057_0_170.json new file mode 100644 index 0000000000000000000000000000000000000000..566512ddc95dca9ea458c9422c30ba952787bad8 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_httbW7TujP0_00057_0_170.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_httbW7TujP0_00057_0_170", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/httbW7TujP0_00057_0_170", + "duration": 5.673 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/httbW7TujP0_00057_0_170_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/httbW7TujP0_00057_0_170_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/httbW7TujP0_00057_0_170", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_iIVJN0Yz1Y0_00027_0_84.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_iIVJN0Yz1Y0_00027_0_84.json new file mode 100644 index 0000000000000000000000000000000000000000..42b646a388728bd0249038abd1fd258ce5c34392 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_iIVJN0Yz1Y0_00027_0_84.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_iIVJN0Yz1Y0_00027_0_84", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/iIVJN0Yz1Y0_00027_0_84", + "duration": 3.4 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/iIVJN0Yz1Y0_00027_0_84_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/iIVJN0Yz1Y0_00027_0_84_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/iIVJN0Yz1Y0_00027_0_84", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_is-juN3Pkfg_00008_0_101.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_is-juN3Pkfg_00008_0_101.json new file mode 100644 index 0000000000000000000000000000000000000000..4aecc5b1789cbcf2620f6cd72766f782dc627c23 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_is-juN3Pkfg_00008_0_101.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_is-juN3Pkfg_00008_0_101", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/is-juN3Pkfg_00008_0_101", + "duration": 4.255 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/is-juN3Pkfg_00008_0_101_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/is-juN3Pkfg_00008_0_101", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ish8GggzGsA_00006_0_109.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ish8GggzGsA_00006_0_109.json new file mode 100644 index 0000000000000000000000000000000000000000..a63882626935036a8249011f30000bf13707b547 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ish8GggzGsA_00006_0_109.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_ish8GggzGsA_00006_0_109", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/ish8GggzGsA_00006_0_109", + "duration": 3.704 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ish8GggzGsA_00006_0_109_joint0_skel.jpg", + "time": 0.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ish8GggzGsA_00006_0_109_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/ish8GggzGsA_00006_0_109", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_j88cdZCpDEM_00008_0_150.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_j88cdZCpDEM_00008_0_150.json new file mode 100644 index 0000000000000000000000000000000000000000..13cdea3ac873b242a36f355823e67a6f27b9a7c4 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_j88cdZCpDEM_00008_0_150.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_j88cdZCpDEM_00008_0_150", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/j88cdZCpDEM_00008_0_150", + "duration": 2.514 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/j88cdZCpDEM_00008_0_150_joint0_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/j88cdZCpDEM_00008_0_150", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jA5Xq4Bs2yE_00065_0_87.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jA5Xq4Bs2yE_00065_0_87.json new file mode 100644 index 0000000000000000000000000000000000000000..be92297d09f1ad53d5e5c324a473a9ef6a666c17 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jA5Xq4Bs2yE_00065_0_87.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_jA5Xq4Bs2yE_00065_0_87", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/jA5Xq4Bs2yE_00065_0_87", + "duration": 2.97 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jA5Xq4Bs2yE_00065_0_87_joint0_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/jA5Xq4Bs2yE_00065_0_87", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jD8dLyj_7-U_00011_0_95.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jD8dLyj_7-U_00011_0_95.json new file mode 100644 index 0000000000000000000000000000000000000000..f36afda342d17c159d40d8213dc12f1357ea57c8 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jD8dLyj_7-U_00011_0_95.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_jD8dLyj_7-U_00011_0_95", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/jD8dLyj_7-U_00011_0_95", + "duration": 3.204 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jD8dLyj_7-U_00011_0_95_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jD8dLyj_7-U_00011_0_95_joint1_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/jD8dLyj_7-U_00011_0_95", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_je-1dG6dBbM_00008_0_114.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_je-1dG6dBbM_00008_0_114.json new file mode 100644 index 0000000000000000000000000000000000000000..81ca94c9827b6b01f901e0ccfd9d765158c6ce98 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_je-1dG6dBbM_00008_0_114.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_je-1dG6dBbM_00008_0_114", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/je-1dG6dBbM_00008_0_114", + "duration": 4.797 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/je-1dG6dBbM_00008_0_114_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/je-1dG6dBbM_00008_0_114_joint1_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/je-1dG6dBbM_00008_0_114_joint2_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/je-1dG6dBbM_00008_0_114", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jfxw2B0uBX0_00031_293_436.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jfxw2B0uBX0_00031_293_436.json new file mode 100644 index 0000000000000000000000000000000000000000..37541579636d227a487a976e20a450df924f5157 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jfxw2B0uBX0_00031_293_436.json @@ -0,0 +1,44 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_jfxw2B0uBX0_00031_293_436", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/jfxw2B0uBX0_00031_293_436", + "duration": 2.41 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jfxw2B0uBX0_00031_293_436_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jfxw2B0uBX0_00031_293_436_joint1_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jfxw2B0uBX0_00031_293_436_joint2_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jfxw2B0uBX0_00031_293_436_joint3_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/jfxw2B0uBX0_00031_293_436", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jhlZItT7mmo_00005_235_320.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jhlZItT7mmo_00005_235_320.json new file mode 100644 index 0000000000000000000000000000000000000000..aac51e86ed721201d114d42d07a8f1bf4067735f --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jhlZItT7mmo_00005_235_320.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_jhlZItT7mmo_00005_235_320", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/jhlZItT7mmo_00005_235_320", + "duration": 2.838 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jhlZItT7mmo_00005_235_320_joint0_skel.jpg", + "time": 0.7 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/jhlZItT7mmo_00005_235_320", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jmw0Kz5K1ms_00014_0_162.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jmw0Kz5K1ms_00014_0_162.json new file mode 100644 index 0000000000000000000000000000000000000000..7faf8c01a4166e6f32ec108c9ed3fb80eb455575 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jmw0Kz5K1ms_00014_0_162.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_jmw0Kz5K1ms_00014_0_162", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/jmw0Kz5K1ms_00014_0_162", + "duration": 5.413 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jmw0Kz5K1ms_00014_0_162_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/jmw0Kz5K1ms_00014_0_162", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jnVb7_CVh6U_00003_173_269.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jnVb7_CVh6U_00003_173_269.json new file mode 100644 index 0000000000000000000000000000000000000000..215867d936610772db429fb59132f975730367ca --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jnVb7_CVh6U_00003_173_269.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_jnVb7_CVh6U_00003_173_269", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/jnVb7_CVh6U_00003_173_269", + "duration": 3.88 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jnVb7_CVh6U_00003_173_269_joint0_skel.jpg", + "time": 0.7 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/jnVb7_CVh6U_00003_173_269", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jqNk4ExKLXI_00033_14_100.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jqNk4ExKLXI_00033_14_100.json new file mode 100644 index 0000000000000000000000000000000000000000..3cefc1bafabdcd0b63c2e5f9572967f03ad728a6 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jqNk4ExKLXI_00033_14_100.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_jqNk4ExKLXI_00033_14_100", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/jqNk4ExKLXI_00033_14_100", + "duration": 3.56 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jqNk4ExKLXI_00033_14_100_joint0_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/jqNk4ExKLXI_00033_14_100", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jz8L6htl0yA_00038_311_415.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jz8L6htl0yA_00038_311_415.json new file mode 100644 index 0000000000000000000000000000000000000000..464be9cabb75139a6f2db705060bbcb65cd1a5c3 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_jz8L6htl0yA_00038_311_415.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_jz8L6htl0yA_00038_311_415", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/jz8L6htl0yA_00038_311_415", + "duration": 3.504 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jz8L6htl0yA_00038_311_415_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/jz8L6htl0yA_00038_311_415_joint1_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/jz8L6htl0yA_00038_311_415", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_k0X8Mz1ksEg_00013_0_154.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_k0X8Mz1ksEg_00013_0_154.json new file mode 100644 index 0000000000000000000000000000000000000000..737bfcfa48cc78eeea1d75e1bf8b2b55a8289b78 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_k0X8Mz1ksEg_00013_0_154.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_k0X8Mz1ksEg_00013_0_154", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/k0X8Mz1ksEg_00013_0_154", + "duration": 6.2 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is rotating right and looking downward, then realigning forward before stooping to retrieve and throw an orange object while advancing.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/k0X8Mz1ksEg_00013_0_154_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/k0X8Mz1ksEg_00013_0_154_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/k0X8Mz1ksEg_00013_0_154", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kHQ8ib3EQb8_00014_0_556.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kHQ8ib3EQb8_00014_0_556.json new file mode 100644 index 0000000000000000000000000000000000000000..2d67fbf43e0c6682f99f8a4fdc2fbfdb5c90233a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kHQ8ib3EQb8_00014_0_556.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_kHQ8ib3EQb8_00014_0_556", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/kHQ8ib3EQb8_00014_0_556", + "duration": 11.14 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kHQ8ib3EQb8_00014_0_556_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kHQ8ib3EQb8_00014_0_556_joint1_skel.jpg", + "time": 6.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kHQ8ib3EQb8_00014_0_556_joint2_skel.jpg", + "time": 10.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/kHQ8ib3EQb8_00014_0_556", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kHU2bS89gVI_00010_178_246.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kHU2bS89gVI_00010_178_246.json new file mode 100644 index 0000000000000000000000000000000000000000..57e4d7d62f7d64ff0f3bc5957750252069583d33 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kHU2bS89gVI_00010_178_246.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_kHU2bS89gVI_00010_178_246", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/kHU2bS89gVI_00010_178_246", + "duration": 2.336 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kHU2bS89gVI_00010_178_246_joint0_skel.jpg", + "time": 0.9 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kHU2bS89gVI_00010_178_246_joint1_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/kHU2bS89gVI_00010_178_246", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kN8xl4KgjNE_00000_29_192.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kN8xl4KgjNE_00000_29_192.json new file mode 100644 index 0000000000000000000000000000000000000000..8b71f3c9411da55c517ff8794a255c6f0a7940ef --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kN8xl4KgjNE_00000_29_192.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_kN8xl4KgjNE_00000_29_192", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/kN8xl4KgjNE_00000_29_192", + "duration": 6.793 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kN8xl4KgjNE_00000_29_192_joint0_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/kN8xl4KgjNE_00000_29_192", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kSm_3Y8jnrA_00007_113_301.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kSm_3Y8jnrA_00007_113_301.json new file mode 100644 index 0000000000000000000000000000000000000000..0871a3ff7a996964b23b16387be4b7c0f22e2dc8 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kSm_3Y8jnrA_00007_113_301.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_kSm_3Y8jnrA_00007_113_301", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/kSm_3Y8jnrA_00007_113_301", + "duration": 3.157 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kSm_3Y8jnrA_00007_113_301_joint0_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/kSm_3Y8jnrA_00007_113_301", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kgmq3A6RzIw_00000_325_677.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kgmq3A6RzIw_00000_325_677.json new file mode 100644 index 0000000000000000000000000000000000000000..d8bc49bf43ad371a70366e23af9dd4bd018c6804 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kgmq3A6RzIw_00000_325_677.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_kgmq3A6RzIw_00000_325_677", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/kgmq3A6RzIw_00000_325_677", + "duration": 11.779 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kgmq3A6RzIw_00000_325_677_joint0_skel.jpg", + "time": 6.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/kgmq3A6RzIw_00000_325_677", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_km7u5M35xb4_00025_0_105.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_km7u5M35xb4_00025_0_105.json new file mode 100644 index 0000000000000000000000000000000000000000..eb6c6ebb15969355f5dc2ac3d9036dae6987bf02 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_km7u5M35xb4_00025_0_105.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_km7u5M35xb4_00025_0_105", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/km7u5M35xb4_00025_0_105", + "duration": 4.24 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/km7u5M35xb4_00025_0_105_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/km7u5M35xb4_00025_0_105", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kvfNBJXHugg_00048_549_824.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kvfNBJXHugg_00048_549_824.json new file mode 100644 index 0000000000000000000000000000000000000000..d8b65325c5dae94b5d012d055a79e9b8c48095bb --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_kvfNBJXHugg_00048_549_824.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_kvfNBJXHugg_00048_549_824", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/kvfNBJXHugg_00048_549_824", + "duration": 9.276 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kvfNBJXHugg_00048_549_824_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/kvfNBJXHugg_00048_549_824_joint1_skel.jpg", + "time": 6.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/kvfNBJXHugg_00048_549_824", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_qqANXkZiVoU_00027_0_254.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_qqANXkZiVoU_00027_0_254.json new file mode 100644 index 0000000000000000000000000000000000000000..dfb997aca99bbfe4179ebd50eedd700c2cb17ecb --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_qqANXkZiVoU_00027_0_254.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_qqANXkZiVoU_00027_0_254", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/qqANXkZiVoU_00027_0_254", + "duration": 4.271 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/qqANXkZiVoU_00027_0_254_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/qqANXkZiVoU_00027_0_254_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/qqANXkZiVoU_00027_0_254", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_qvQSt_eqISY_00002_103_188.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_qvQSt_eqISY_00002_103_188.json new file mode 100644 index 0000000000000000000000000000000000000000..302c7d58129b73967e6e098c3a9fcd20e6a35852 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_qvQSt_eqISY_00002_103_188.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_qvQSt_eqISY_00002_103_188", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/qvQSt_eqISY_00002_103_188", + "duration": 2.87 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/qvQSt_eqISY_00002_103_188_joint0_skel.jpg", + "time": 0.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/qvQSt_eqISY_00002_103_188", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_r8gxkZynhvk_00008_0_104.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_r8gxkZynhvk_00008_0_104.json new file mode 100644 index 0000000000000000000000000000000000000000..f44c45c09a3ba7058cc21eacd221ae347cb22b2d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_r8gxkZynhvk_00008_0_104.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_r8gxkZynhvk_00008_0_104", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/r8gxkZynhvk_00008_0_104", + "duration": 3.537 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/r8gxkZynhvk_00008_0_104_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/r8gxkZynhvk_00008_0_104", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rEtmuByf-0E_00006_1922_2189.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rEtmuByf-0E_00006_1922_2189.json new file mode 100644 index 0000000000000000000000000000000000000000..763e735732ae8144eae2313f5177685ba6463c07 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rEtmuByf-0E_00006_1922_2189.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_rEtmuByf-0E_00006_1922_2189", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/rEtmuByf-0E_00006_1922_2189", + "duration": 5.4 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rEtmuByf-0E_00006_1922_2189_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rEtmuByf-0E_00006_1922_2189_joint1_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/rEtmuByf-0E_00006_1922_2189", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rGsF1notX2A_00020_0_88.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rGsF1notX2A_00020_0_88.json new file mode 100644 index 0000000000000000000000000000000000000000..3c95dd3db0618ee177364097b9052270e3bfe3af --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rGsF1notX2A_00020_0_88.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_rGsF1notX2A_00020_0_88", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/rGsF1notX2A_00020_0_88", + "duration": 3.003 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rGsF1notX2A_00020_0_88_joint0_skel.jpg", + "time": 0.8 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rGsF1notX2A_00020_0_88_joint1_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/rGsF1notX2A_00020_0_88", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rLoHXD9dxUU_00042_97_166.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rLoHXD9dxUU_00042_97_166.json new file mode 100644 index 0000000000000000000000000000000000000000..8a30bbbf3b32baaa3e475b1af29e005a22ab3f06 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rLoHXD9dxUU_00042_97_166.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_rLoHXD9dxUU_00042_97_166", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/rLoHXD9dxUU_00042_97_166", + "duration": 2.917 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rLoHXD9dxUU_00042_97_166_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rLoHXD9dxUU_00042_97_166_joint1_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/rLoHXD9dxUU_00042_97_166", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rPvkdU5kkqA_00010_0_59.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rPvkdU5kkqA_00010_0_59.json new file mode 100644 index 0000000000000000000000000000000000000000..90c19a422de502c650d5d3ed61aee56b26d53859 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rPvkdU5kkqA_00010_0_59.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_rPvkdU5kkqA_00010_0_59", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/rPvkdU5kkqA_00010_0_59", + "duration": 2.002 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rPvkdU5kkqA_00010_0_59_joint0_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/rPvkdU5kkqA_00010_0_59", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rVq_-sLnHY4_00004_0_76.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rVq_-sLnHY4_00004_0_76.json new file mode 100644 index 0000000000000000000000000000000000000000..405cc56b1ce193e1dd0a7a1b8851e04e2cd8454a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rVq_-sLnHY4_00004_0_76.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_rVq_-sLnHY4_00004_0_76", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/rVq_-sLnHY4_00004_0_76", + "duration": 2.57 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rVq_-sLnHY4_00004_0_76_joint0_skel.jpg", + "time": 1.7 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/rVq_-sLnHY4_00004_0_76", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_re5DQem4pGg_00001_0_68.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_re5DQem4pGg_00001_0_68.json new file mode 100644 index 0000000000000000000000000000000000000000..a2abe8988d9e712a407509273ddf7e9a038c28da --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_re5DQem4pGg_00001_0_68.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_re5DQem4pGg_00001_0_68", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/re5DQem4pGg_00001_0_68", + "duration": 2.37 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/re5DQem4pGg_00001_0_68_joint0_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/re5DQem4pGg_00001_0_68", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rguBtTP7mw4_00022_0_406.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rguBtTP7mw4_00022_0_406.json new file mode 100644 index 0000000000000000000000000000000000000000..f3819659bbddb17b885cf5c9920d1647f7e9de5a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rguBtTP7mw4_00022_0_406.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_rguBtTP7mw4_00022_0_406", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/rguBtTP7mw4_00022_0_406", + "duration": 6.817 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rguBtTP7mw4_00022_0_406_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rguBtTP7mw4_00022_0_406_joint1_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/rguBtTP7mw4_00022_0_406", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rz1r2f68qx8_00002_220_296.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rz1r2f68qx8_00002_220_296.json new file mode 100644 index 0000000000000000000000000000000000000000..9531bce4db54a9f6c431f30359dfad0ef7e32369 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_rz1r2f68qx8_00002_220_296.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_rz1r2f68qx8_00002_220_296", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/rz1r2f68qx8_00002_220_296", + "duration": 3.175 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rz1r2f68qx8_00002_220_296_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/rz1r2f68qx8_00002_220_296_joint1_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/rz1r2f68qx8_00002_220_296", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s3L1T34Zrdo_00047_0_236.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s3L1T34Zrdo_00047_0_236.json new file mode 100644 index 0000000000000000000000000000000000000000..a5303b38b3b5b0ea43eba01315c385a538e43b82 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s3L1T34Zrdo_00047_0_236.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_s3L1T34Zrdo_00047_0_236", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/s3L1T34Zrdo_00047_0_236", + "duration": 7.913 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/s3L1T34Zrdo_00047_0_236_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/s3L1T34Zrdo_00047_0_236_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/s3L1T34Zrdo_00047_0_236", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s3PvzlrzuQ4_00017_83_141.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s3PvzlrzuQ4_00017_83_141.json new file mode 100644 index 0000000000000000000000000000000000000000..d71ec780daaad8fb9ebe82b7714c2418dca4da19 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s3PvzlrzuQ4_00017_83_141.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_s3PvzlrzuQ4_00017_83_141", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/s3PvzlrzuQ4_00017_83_141", + "duration": 2.545 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/s3PvzlrzuQ4_00017_83_141_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/s3PvzlrzuQ4_00017_83_141_joint1_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/s3PvzlrzuQ4_00017_83_141_joint2_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/s3PvzlrzuQ4_00017_83_141", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s5x2LY5TRv8_00027_0_152.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s5x2LY5TRv8_00027_0_152.json new file mode 100644 index 0000000000000000000000000000000000000000..88e0e08f36c6e7cad867737683c87c4bf3c5b009 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_s5x2LY5TRv8_00027_0_152.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_s5x2LY5TRv8_00027_0_152", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/s5x2LY5TRv8_00027_0_152", + "duration": 6.341 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/s5x2LY5TRv8_00027_0_152_joint0_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/s5x2LY5TRv8_00027_0_152_joint1_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/s5x2LY5TRv8_00027_0_152", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sCOmqB73HSE_00007_984_1132.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sCOmqB73HSE_00007_984_1132.json new file mode 100644 index 0000000000000000000000000000000000000000..c46cb38f633abfa1dce61df6a4ddae1cc90d8d0c --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sCOmqB73HSE_00007_984_1132.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_sCOmqB73HSE_00007_984_1132", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/sCOmqB73HSE_00007_984_1132", + "duration": 5.072 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/sCOmqB73HSE_00007_984_1132_joint0_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/sCOmqB73HSE_00007_984_1132", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sGr1HBSIic8_00022_242_384.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sGr1HBSIic8_00022_242_384.json new file mode 100644 index 0000000000000000000000000000000000000000..e9a9362b3875decb50aaf93fe749d7c3351e8f6c --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sGr1HBSIic8_00022_242_384.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_sGr1HBSIic8_00022_242_384", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/sGr1HBSIic8_00022_242_384", + "duration": 4.738 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/sGr1HBSIic8_00022_242_384_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/sGr1HBSIic8_00022_242_384", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sTpqVab63mY_00025_0_64.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sTpqVab63mY_00025_0_64.json new file mode 100644 index 0000000000000000000000000000000000000000..701caddefbe28ed67e40347093785512b0c56afc --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sTpqVab63mY_00025_0_64.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_sTpqVab63mY_00025_0_64", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/sTpqVab63mY_00025_0_64", + "duration": 2.56 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is engaging with material at a table and then moves purposefully toward a different location to begin another task.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/sTpqVab63mY_00025_0_64_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/sTpqVab63mY_00025_0_64", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sdRHReep4mw_00018_0_70.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sdRHReep4mw_00018_0_70.json new file mode 100644 index 0000000000000000000000000000000000000000..57e77d551bf15e8d34a574bb4e817648e984501e --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sdRHReep4mw_00018_0_70.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_sdRHReep4mw_00018_0_70", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/sdRHReep4mw_00018_0_70", + "duration": 2.961 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/sdRHReep4mw_00018_0_70_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/sdRHReep4mw_00018_0_70", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sfOi1AdSkjk_00011_0_127.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sfOi1AdSkjk_00011_0_127.json new file mode 100644 index 0000000000000000000000000000000000000000..fd72dd5cdbf258b43c5ab00b1f7738d9c4b371f5 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sfOi1AdSkjk_00011_0_127.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_sfOi1AdSkjk_00011_0_127", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/sfOi1AdSkjk_00011_0_127", + "duration": 5.339 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/sfOi1AdSkjk_00011_0_127_joint0_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/sfOi1AdSkjk_00011_0_127", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_slCwOLXIeU4_00025_56_104.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_slCwOLXIeU4_00025_56_104.json new file mode 100644 index 0000000000000000000000000000000000000000..fa3f4c9aa3350d57d331a24bcda9f0f26981c3ec --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_slCwOLXIeU4_00025_56_104.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_slCwOLXIeU4_00025_56_104", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/slCwOLXIeU4_00025_56_104", + "duration": 2.044 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/slCwOLXIeU4_00025_56_104_joint0_skel.jpg", + "time": 0.4 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/slCwOLXIeU4_00025_56_104_joint1_skel.jpg", + "time": 0.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/slCwOLXIeU4_00025_56_104", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_slcf7Y3yEdU_00037_0_411.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_slcf7Y3yEdU_00037_0_411.json new file mode 100644 index 0000000000000000000000000000000000000000..317d45460eb077a9a2dc41d3af462c16f626f6af --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_slcf7Y3yEdU_00037_0_411.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_slcf7Y3yEdU_00037_0_411", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/slcf7Y3yEdU_00037_0_411", + "duration": 8.233 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/slcf7Y3yEdU_00037_0_411_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/slcf7Y3yEdU_00037_0_411_joint1_skel.jpg", + "time": 5.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/slcf7Y3yEdU_00037_0_411", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_svu306fZ5w8_00015_0_72.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_svu306fZ5w8_00015_0_72.json new file mode 100644 index 0000000000000000000000000000000000000000..db9ab174cf8364fb0e9f72b72d06a6eb2cb0ba63 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_svu306fZ5w8_00015_0_72.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_svu306fZ5w8_00015_0_72", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/svu306fZ5w8_00015_0_72", + "duration": 2.96 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/svu306fZ5w8_00015_0_72_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/svu306fZ5w8_00015_0_72", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sxJh8fI35iw_00033_0_51.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sxJh8fI35iw_00033_0_51.json new file mode 100644 index 0000000000000000000000000000000000000000..a19f13e66d5a9203f948ce33d7277d7c45f48203 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_sxJh8fI35iw_00033_0_51.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_sxJh8fI35iw_00033_0_51", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/sxJh8fI35iw_00033_0_51", + "duration": 2.169 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is standing in a kitchen and speaking toward the camera before leaning toward a counter to transfer the contents of a glass bowl into a pot.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/sxJh8fI35iw_00033_0_51_joint0_skel.jpg", + "time": 0.7 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/sxJh8fI35iw_00033_0_51", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t3JoimGNERA_00010_0_88.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t3JoimGNERA_00010_0_88.json new file mode 100644 index 0000000000000000000000000000000000000000..a1c321480f1665b45b35fc083a8341e8ad5e949a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t3JoimGNERA_00010_0_88.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_t3JoimGNERA_00010_0_88", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/t3JoimGNERA_00010_0_88", + "duration": 3.003 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/t3JoimGNERA_00010_0_88_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/t3JoimGNERA_00010_0_88_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/t3JoimGNERA_00010_0_88", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t3ybHGZxjjQ_00014_0_208.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t3ybHGZxjjQ_00014_0_208.json new file mode 100644 index 0000000000000000000000000000000000000000..bd83390c6abf8cc91ddd9d05a1d42e4d768998f6 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t3ybHGZxjjQ_00014_0_208.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_t3ybHGZxjjQ_00014_0_208", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/t3ybHGZxjjQ_00014_0_208", + "duration": 8.676 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/t3ybHGZxjjQ_00014_0_208_joint0_skel.jpg", + "time": 2.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/t3ybHGZxjjQ_00014_0_208_joint1_skel.jpg", + "time": 5.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/t3ybHGZxjjQ_00014_0_208", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t5wg4c0ScAc_00035_453_724.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t5wg4c0ScAc_00035_453_724.json new file mode 100644 index 0000000000000000000000000000000000000000..50e257ec64f7618eae820f178f8e7cdd77ee968e --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_t5wg4c0ScAc_00035_453_724.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_t5wg4c0ScAc_00035_453_724", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/t5wg4c0ScAc_00035_453_724", + "duration": 5.44 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/t5wg4c0ScAc_00035_453_724_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/t5wg4c0ScAc_00035_453_724_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/t5wg4c0ScAc_00035_453_724", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tFc5jBcitPY_00021_54_214.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tFc5jBcitPY_00021_54_214.json new file mode 100644 index 0000000000000000000000000000000000000000..8a7dac4d94425988bf9cbbcffd619fcd16b6be90 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tFc5jBcitPY_00021_54_214.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tFc5jBcitPY_00021_54_214", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tFc5jBcitPY_00021_54_214", + "duration": 6.4 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tFc5jBcitPY_00021_54_214_joint0_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tFc5jBcitPY_00021_54_214_joint1_skel.jpg", + "time": 3.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tFc5jBcitPY_00021_54_214", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tHokJbz14nA_00024_0_240.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tHokJbz14nA_00024_0_240.json new file mode 100644 index 0000000000000000000000000000000000000000..5062e0f79eaf5fe40bf49fd6e363794b41d427da --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tHokJbz14nA_00024_0_240.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tHokJbz14nA_00024_0_240", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tHokJbz14nA_00024_0_240", + "duration": 4.038 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tHokJbz14nA_00024_0_240_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tHokJbz14nA_00024_0_240_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tHokJbz14nA_00024_0_240", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tLnTZliGbAc_00072_0_147.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tLnTZliGbAc_00072_0_147.json new file mode 100644 index 0000000000000000000000000000000000000000..b590869beadcad30bf24dc314a34248a649ea0c1 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tLnTZliGbAc_00072_0_147.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tLnTZliGbAc_00072_0_147", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tLnTZliGbAc_00072_0_147", + "duration": 4.939 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tLnTZliGbAc_00072_0_147_joint0_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tLnTZliGbAc_00072_0_147", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tMSHEtgk4xA_00009_74_168.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tMSHEtgk4xA_00009_74_168.json new file mode 100644 index 0000000000000000000000000000000000000000..c95967a2bdbb7f1c487295d4ddf85e524285810d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tMSHEtgk4xA_00009_74_168.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tMSHEtgk4xA_00009_74_168", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tMSHEtgk4xA_00009_74_168", + "duration": 3.17 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tMSHEtgk4xA_00009_74_168_joint0_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tMSHEtgk4xA_00009_74_168", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tUjWaZMz_Yo_00014_0_63.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tUjWaZMz_Yo_00014_0_63.json new file mode 100644 index 0000000000000000000000000000000000000000..c3e68abdb92b60f89bffae07fdc2dff4ae199712 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tUjWaZMz_Yo_00014_0_63.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tUjWaZMz_Yo_00014_0_63", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tUjWaZMz_Yo_00014_0_63", + "duration": 2.64 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tUjWaZMz_Yo_00014_0_63_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tUjWaZMz_Yo_00014_0_63", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tYi23Gsg6Is_00017_157_465.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tYi23Gsg6Is_00017_157_465.json new file mode 100644 index 0000000000000000000000000000000000000000..945db29e537e57f3fb5cb8cbce6fd8f0a9789208 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tYi23Gsg6Is_00017_157_465.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tYi23Gsg6Is_00017_157_465", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tYi23Gsg6Is_00017_157_465", + "duration": 10.378 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tYi23Gsg6Is_00017_157_465_joint0_skel.jpg", + "time": 2.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tYi23Gsg6Is_00017_157_465_joint1_skel.jpg", + "time": 5.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tYi23Gsg6Is_00017_157_465_joint2_skel.jpg", + "time": 8.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tYi23Gsg6Is_00017_157_465", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tYnPjJD8Y0I_00072_231_303.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tYnPjJD8Y0I_00072_231_303.json new file mode 100644 index 0000000000000000000000000000000000000000..36955b3c3815aa2386db038bd8889ffd84c8524d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tYnPjJD8Y0I_00072_231_303.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tYnPjJD8Y0I_00072_231_303", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tYnPjJD8Y0I_00072_231_303", + "duration": 2.96 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tYnPjJD8Y0I_00072_231_303_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tYnPjJD8Y0I_00072_231_303_joint1_skel.jpg", + "time": 1.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tYnPjJD8Y0I_00072_231_303", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tcUJC5EVwRk_00035_0_63.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tcUJC5EVwRk_00035_0_63.json new file mode 100644 index 0000000000000000000000000000000000000000..d8304e2765aeb156a8b6657f870f6eaaed020bdd --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tcUJC5EVwRk_00035_0_63.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tcUJC5EVwRk_00035_0_63", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tcUJC5EVwRk_00035_0_63", + "duration": 2.236 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tcUJC5EVwRk_00035_0_63_joint0_skel.jpg", + "time": 0.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tcUJC5EVwRk_00035_0_63_joint1_skel.jpg", + "time": 1.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tcUJC5EVwRk_00035_0_63", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tevz58dPHXw_00000_0_91.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tevz58dPHXw_00000_0_91.json new file mode 100644 index 0000000000000000000000000000000000000000..7bddf7af917bd98d107d4a7ed14b4dc99c96f66f --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tevz58dPHXw_00000_0_91.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tevz58dPHXw_00000_0_91", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tevz58dPHXw_00000_0_91", + "duration": 3.204 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tevz58dPHXw_00000_0_91_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tevz58dPHXw_00000_0_91", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tfXFKIQ10ro_00023_0_315.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tfXFKIQ10ro_00023_0_315.json new file mode 100644 index 0000000000000000000000000000000000000000..539588dce801fe77b6f59951cadd2a44f520a715 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tfXFKIQ10ro_00023_0_315.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tfXFKIQ10ro_00023_0_315", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tfXFKIQ10ro_00023_0_315", + "duration": 5.281 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tfXFKIQ10ro_00023_0_315_joint0_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tfXFKIQ10ro_00023_0_315_joint1_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tfXFKIQ10ro_00023_0_315", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tkc8RnGP86w_00006_484_1006.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tkc8RnGP86w_00006_484_1006.json new file mode 100644 index 0000000000000000000000000000000000000000..b0ccfa36349a500383f51fe43238cbec8f5121f7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_tkc8RnGP86w_00006_484_1006.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_tkc8RnGP86w_00006_484_1006", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/tkc8RnGP86w_00006_484_1006", + "duration": 17.434 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tkc8RnGP86w_00006_484_1006_joint0_skel.jpg", + "time": 3.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tkc8RnGP86w_00006_484_1006_joint1_skel.jpg", + "time": 7.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/tkc8RnGP86w_00006_484_1006_joint2_skel.jpg", + "time": 10.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/tkc8RnGP86w_00006_484_1006", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_todl3gWNdag_00004_0_118.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_todl3gWNdag_00004_0_118.json new file mode 100644 index 0000000000000000000000000000000000000000..45666b58c14f2da311857c8fe8d0699740d0bd9e --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_todl3gWNdag_00004_0_118.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_todl3gWNdag_00004_0_118", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/todl3gWNdag_00004_0_118", + "duration": 4.88 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is standing at a kitchen counter, gesturing as if explaining something, then turning to retrieve a paper towel from a dispenser before facing forward again.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/todl3gWNdag_00004_0_118_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/todl3gWNdag_00004_0_118", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_u-27fhZvFrU_00038_0_210.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_u-27fhZvFrU_00038_0_210.json new file mode 100644 index 0000000000000000000000000000000000000000..89e0dcc82f43e86d3a22f16493c2ecdf94987313 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_u-27fhZvFrU_00038_0_210.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_u-27fhZvFrU_00038_0_210", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/u-27fhZvFrU_00038_0_210", + "duration": 3.522 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/u-27fhZvFrU_00038_0_210_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/u-27fhZvFrU_00038_0_210_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/u-27fhZvFrU_00038_0_210", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_u1PVDPE-hgE_00007_0_86.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_u1PVDPE-hgE_00007_0_86.json new file mode 100644 index 0000000000000000000000000000000000000000..0be1bf88bf48aa9a73d4056acf751a4d34d0b24d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_u1PVDPE-hgE_00007_0_86.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_u1PVDPE-hgE_00007_0_86", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/u1PVDPE-hgE_00007_0_86", + "duration": 3.629 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/u1PVDPE-hgE_00007_0_86_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/u1PVDPE-hgE_00007_0_86", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uQpxQUE5XiA_00000_634_1022.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uQpxQUE5XiA_00000_634_1022.json new file mode 100644 index 0000000000000000000000000000000000000000..07fbfc2c13db613432678a82249f2f94e79edb7e --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uQpxQUE5XiA_00000_634_1022.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_uQpxQUE5XiA_00000_634_1022", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/uQpxQUE5XiA_00000_634_1022", + "duration": 12.98 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/uQpxQUE5XiA_00000_634_1022_joint0_skel.jpg", + "time": 2.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/uQpxQUE5XiA_00000_634_1022_joint1_skel.jpg", + "time": 6.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/uQpxQUE5XiA_00000_634_1022_joint2_skel.jpg", + "time": 10.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/uQpxQUE5XiA_00000_634_1022", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uYBMmOeixow_00040_185_304.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uYBMmOeixow_00040_185_304.json new file mode 100644 index 0000000000000000000000000000000000000000..3cd0cee6fe94cbcbdbc94f02cd0f67f858fdaeda --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uYBMmOeixow_00040_185_304.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_uYBMmOeixow_00040_185_304", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/uYBMmOeixow_00040_185_304", + "duration": 3.95 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/uYBMmOeixow_00040_185_304_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/uYBMmOeixow_00040_185_304", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uao897AZC08_00024_64_161.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uao897AZC08_00024_64_161.json new file mode 100644 index 0000000000000000000000000000000000000000..8e90f44902175a989cdfde7968a14d9bd28a7d41 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uao897AZC08_00024_64_161.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_uao897AZC08_00024_64_161", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/uao897AZC08_00024_64_161", + "duration": 3.27 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/uao897AZC08_00024_64_161_joint0_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/uao897AZC08_00024_64_161", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uhuO0DFe0ZY_00033_0_114.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uhuO0DFe0ZY_00033_0_114.json new file mode 100644 index 0000000000000000000000000000000000000000..644453f822d6f0753dc3ffd3f342ffae81fca4bd --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uhuO0DFe0ZY_00033_0_114.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_uhuO0DFe0ZY_00033_0_114", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/uhuO0DFe0ZY_00033_0_114", + "duration": 4.6 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/uhuO0DFe0ZY_00033_0_114_joint0_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/uhuO0DFe0ZY_00033_0_114", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uieRXHzxCRQ_00041_144_324.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uieRXHzxCRQ_00041_144_324.json new file mode 100644 index 0000000000000000000000000000000000000000..e2e0f16e8a450a6b83c6749c658f87d44d3032c1 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_uieRXHzxCRQ_00041_144_324.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_uieRXHzxCRQ_00041_144_324", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/uieRXHzxCRQ_00041_144_324", + "duration": 6.04 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/uieRXHzxCRQ_00041_144_324_joint0_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/uieRXHzxCRQ_00041_144_324", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ux467aEJkjo_00035_0_182.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ux467aEJkjo_00035_0_182.json new file mode 100644 index 0000000000000000000000000000000000000000..ff3f9078d74d4d2d3fa060a7017a4f99465fb386 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_ux467aEJkjo_00035_0_182.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_ux467aEJkjo_00035_0_182", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/ux467aEJkjo_00035_0_182", + "duration": 3.054 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ux467aEJkjo_00035_0_182_joint0_skel.jpg", + "time": 1.2 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/ux467aEJkjo_00035_0_182_joint1_skel.jpg", + "time": 2.7 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/ux467aEJkjo_00035_0_182", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v3J14lxJmHo_00021_0_178.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v3J14lxJmHo_00021_0_178.json new file mode 100644 index 0000000000000000000000000000000000000000..ddf8c756211144a230c27ac60e143b21765c13e7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v3J14lxJmHo_00021_0_178.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_v3J14lxJmHo_00021_0_178", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/v3J14lxJmHo_00021_0_178", + "duration": 2.997 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/v3J14lxJmHo_00021_0_178_joint0_skel.jpg", + "time": 0.4 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/v3J14lxJmHo_00021_0_178_joint1_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/v3J14lxJmHo_00021_0_178", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v5dFCq3ZZlo_00014_482_620.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v5dFCq3ZZlo_00014_482_620.json new file mode 100644 index 0000000000000000000000000000000000000000..3de8a43d3a9be44057569b1c8393db8d3edf7915 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v5dFCq3ZZlo_00014_482_620.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_v5dFCq3ZZlo_00014_482_620", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/v5dFCq3ZZlo_00014_482_620", + "duration": 2.321 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/v5dFCq3ZZlo_00014_482_620_joint0_skel.jpg", + "time": 1.3 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/v5dFCq3ZZlo_00014_482_620", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v81_5bBEakI_00085_0_247.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v81_5bBEakI_00085_0_247.json new file mode 100644 index 0000000000000000000000000000000000000000..df7200ba24c3a8fb3a9e1be354c3ef8252f6dbd7 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v81_5bBEakI_00085_0_247.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_v81_5bBEakI_00085_0_247", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/v81_5bBEakI_00085_0_247", + "duration": 4.98 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/v81_5bBEakI_00085_0_247_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/v81_5bBEakI_00085_0_247_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/v81_5bBEakI_00085_0_247", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v9m0cWjtP0A_00036_28_148.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v9m0cWjtP0A_00036_28_148.json new file mode 100644 index 0000000000000000000000000000000000000000..912978a29e4be1f58ffe0a4eef1ad6041b2175ba --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_v9m0cWjtP0A_00036_28_148.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_v9m0cWjtP0A_00036_28_148", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/v9m0cWjtP0A_00036_28_148", + "duration": 4.007 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/v9m0cWjtP0A_00036_28_148_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/v9m0cWjtP0A_00036_28_148", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vKREmiZfeCs_00043_0_126.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vKREmiZfeCs_00043_0_126.json new file mode 100644 index 0000000000000000000000000000000000000000..196fb26f1c765b466763b105f0167d80e4457995 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vKREmiZfeCs_00043_0_126.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_vKREmiZfeCs_00043_0_126", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/vKREmiZfeCs_00043_0_126", + "duration": 5.16 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vKREmiZfeCs_00043_0_126_joint0_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/vKREmiZfeCs_00043_0_126", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vOBy-BKZIJc_00008_35_109.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vOBy-BKZIJc_00008_35_109.json new file mode 100644 index 0000000000000000000000000000000000000000..92969cb2a3040068280398c8a15623cf30a040b0 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vOBy-BKZIJc_00008_35_109.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_vOBy-BKZIJc_00008_35_109", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/vOBy-BKZIJc_00008_35_109", + "duration": 2.503 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vOBy-BKZIJc_00008_35_109_joint0_skel.jpg", + "time": 0.8 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vOBy-BKZIJc_00008_35_109_joint1_skel.jpg", + "time": 1.6 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/vOBy-BKZIJc_00008_35_109", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vOzStTFIJ1g_00047_286_460.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vOzStTFIJ1g_00047_286_460.json new file mode 100644 index 0000000000000000000000000000000000000000..925b46d0334477a13598ab6f7b593f146d9bb221 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vOzStTFIJ1g_00047_286_460.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_vOzStTFIJ1g_00047_286_460", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/vOzStTFIJ1g_00047_286_460", + "duration": 2.92 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vOzStTFIJ1g_00047_286_460_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vOzStTFIJ1g_00047_286_460_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/vOzStTFIJ1g_00047_286_460", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vXgqnnkqNok_00008_0_106.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vXgqnnkqNok_00008_0_106.json new file mode 100644 index 0000000000000000000000000000000000000000..426fe224bd98324b0efbd154874864efd7468eeb --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vXgqnnkqNok_00008_0_106.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_vXgqnnkqNok_00008_0_106", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/vXgqnnkqNok_00008_0_106", + "duration": 4.463 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vXgqnnkqNok_00008_0_106_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vXgqnnkqNok_00008_0_106_joint1_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vXgqnnkqNok_00008_0_106_joint2_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/vXgqnnkqNok_00008_0_106", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vbf2cZMFasU_00009_0_66.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vbf2cZMFasU_00009_0_66.json new file mode 100644 index 0000000000000000000000000000000000000000..3610f8b1eae132589b84ebdffb6e5a544ad45968 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vbf2cZMFasU_00009_0_66.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_vbf2cZMFasU_00009_0_66", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/vbf2cZMFasU_00009_0_66", + "duration": 2.303 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vbf2cZMFasU_00009_0_66_joint0_skel.jpg", + "time": 0.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vbf2cZMFasU_00009_0_66_joint1_skel.jpg", + "time": 1.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/vbf2cZMFasU_00009_0_66", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_voMV3Mhta1U_00016_0_106.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_voMV3Mhta1U_00016_0_106.json new file mode 100644 index 0000000000000000000000000000000000000000..d747a707547099ed4485bdd43cd9558313e1dc6a --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_voMV3Mhta1U_00016_0_106.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_voMV3Mhta1U_00016_0_106", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/voMV3Mhta1U_00016_0_106", + "duration": 4.463 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/voMV3Mhta1U_00016_0_106_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/voMV3Mhta1U_00016_0_106_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/voMV3Mhta1U_00016_0_106", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vosAVclG2TI_00000_313_612.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vosAVclG2TI_00000_313_612.json new file mode 100644 index 0000000000000000000000000000000000000000..594bb54355a028e68865dc05a41be4b52df2ef8d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_vosAVclG2TI_00000_313_612.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_vosAVclG2TI_00000_313_612", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/vosAVclG2TI_00000_313_612", + "duration": 10.044 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vosAVclG2TI_00000_313_612_joint0_skel.jpg", + "time": 3.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/vosAVclG2TI_00000_313_612_joint1_skel.jpg", + "time": 6.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/vosAVclG2TI_00000_313_612", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wKH4hW8VQa8_00089_0_221.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wKH4hW8VQa8_00089_0_221.json new file mode 100644 index 0000000000000000000000000000000000000000..12b9a1ce2ce6d7882a48024e5aaa56ed3ee327e0 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wKH4hW8VQa8_00089_0_221.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_wKH4hW8VQa8_00089_0_221", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/wKH4hW8VQa8_00089_0_221", + "duration": 7.408 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wKH4hW8VQa8_00089_0_221_joint0_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/wKH4hW8VQa8_00089_0_221", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wadfKQg4Yt4_00024_0_140.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wadfKQg4Yt4_00024_0_140.json new file mode 100644 index 0000000000000000000000000000000000000000..dbc2f33b8c41dc8030753f6a89fa78129352860d --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wadfKQg4Yt4_00024_0_140.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_wadfKQg4Yt4_00024_0_140", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/wadfKQg4Yt4_00024_0_140", + "duration": 5.881 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is positioned behind a counter and then rotates to perform a playful gesture while moving forward.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wadfKQg4Yt4_00024_0_140_joint0_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/wadfKQg4Yt4_00024_0_140", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wo1yMzZCMko_00002_59_146.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wo1yMzZCMko_00002_59_146.json new file mode 100644 index 0000000000000000000000000000000000000000..2017aa77d9d1532905016816c3b32f17d2ab4e24 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wo1yMzZCMko_00002_59_146.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_wo1yMzZCMko_00002_59_146", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/wo1yMzZCMko_00002_59_146", + "duration": 3.003 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wo1yMzZCMko_00002_59_146_joint0_skel.jpg", + "time": 1.7 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/wo1yMzZCMko_00002_59_146", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wpstnTGr73M_00001_269_432.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wpstnTGr73M_00001_269_432.json new file mode 100644 index 0000000000000000000000000000000000000000..2c4ca52e3e36ee770977183d61a4505066de0a92 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wpstnTGr73M_00001_269_432.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_wpstnTGr73M_00001_269_432", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/wpstnTGr73M_00001_269_432", + "duration": 5.506 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wpstnTGr73M_00001_269_432_joint0_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/wpstnTGr73M_00001_269_432", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wqrpka69v1w_00022_331_483.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wqrpka69v1w_00022_331_483.json new file mode 100644 index 0000000000000000000000000000000000000000..061f7441516d133f84a4891d7dc8c37a377fc4ce --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wqrpka69v1w_00022_331_483.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_wqrpka69v1w_00022_331_483", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/wqrpka69v1w_00022_331_483", + "duration": 5.172 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wqrpka69v1w_00022_331_483_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wqrpka69v1w_00022_331_483_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/wqrpka69v1w_00022_331_483", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wrb-EABXqCY_00029_0_147.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wrb-EABXqCY_00029_0_147.json new file mode 100644 index 0000000000000000000000000000000000000000..d6c27420fe81a409a20708a9dfed865de23e570c --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wrb-EABXqCY_00029_0_147.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_wrb-EABXqCY_00029_0_147", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/wrb-EABXqCY_00029_0_147", + "duration": 2.486 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wrb-EABXqCY_00029_0_147_joint0_skel.jpg", + "time": 1.2 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/wrb-EABXqCY_00029_0_147", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wt3UQzkXcEA_00031_1197_1444.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wt3UQzkXcEA_00031_1197_1444.json new file mode 100644 index 0000000000000000000000000000000000000000..7653e351943647c7ae799c587a9a089c865ad5ba --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wt3UQzkXcEA_00031_1197_1444.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_wt3UQzkXcEA_00031_1197_1444", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/wt3UQzkXcEA_00031_1197_1444", + "duration": 4.143 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wt3UQzkXcEA_00031_1197_1444_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/wt3UQzkXcEA_00031_1197_1444", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wzX4jdw5u1k_00025_0_54.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wzX4jdw5u1k_00025_0_54.json new file mode 100644 index 0000000000000000000000000000000000000000..a038b419dda378d34b626d7d149525218de099de --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_wzX4jdw5u1k_00025_0_54.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_wzX4jdw5u1k_00025_0_54", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/wzX4jdw5u1k_00025_0_54", + "duration": 2.294 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is hanging a pink jacket in a closet, adjusting its position among the other garments, and then walking away.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wzX4jdw5u1k_00025_0_54_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/wzX4jdw5u1k_00025_0_54_joint1_skel.jpg", + "time": 1.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/wzX4jdw5u1k_00025_0_54", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x3jLXMIdPXo_00010_142_271.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x3jLXMIdPXo_00010_142_271.json new file mode 100644 index 0000000000000000000000000000000000000000..72ec18de6b19f9c5a682d3c4f1aa90d51aae99ea --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x3jLXMIdPXo_00010_142_271.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_x3jLXMIdPXo_00010_142_271", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/x3jLXMIdPXo_00010_142_271", + "duration": 4.483 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is walking steadily along a paved outdoor path from left to right before slowing down and turning to approach a black sedan in preparation for entry.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/x3jLXMIdPXo_00010_142_271_joint0_skel.jpg", + "time": 3.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/x3jLXMIdPXo_00010_142_271", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x7yA3xtWNCU_00020_0_105.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x7yA3xtWNCU_00020_0_105.json new file mode 100644 index 0000000000000000000000000000000000000000..545e2d5bd3bf703be1371ec878cf87de7a1cca87 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x7yA3xtWNCU_00020_0_105.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_x7yA3xtWNCU_00020_0_105", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/x7yA3xtWNCU_00020_0_105", + "duration": 4.422 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/x7yA3xtWNCU_00020_0_105_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/x7yA3xtWNCU_00020_0_105_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/x7yA3xtWNCU_00020_0_105", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xD1ywMiX99s_00007_0_194.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xD1ywMiX99s_00007_0_194.json new file mode 100644 index 0000000000000000000000000000000000000000..c781255766eac70a5390e4494c988f64c2f7e04b --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xD1ywMiX99s_00007_0_194.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_xD1ywMiX99s_00007_0_194", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/xD1ywMiX99s_00007_0_194", + "duration": 6.474 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xD1ywMiX99s_00007_0_194_joint0_skel.jpg", + "time": 2.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xD1ywMiX99s_00007_0_194_joint1_skel.jpg", + "time": 4.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/xD1ywMiX99s_00007_0_194", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xD1ywMiX99s_00010_333_755.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xD1ywMiX99s_00010_333_755.json new file mode 100644 index 0000000000000000000000000000000000000000..3f732cb4a25cce6bac2133766249dbef114dd8c3 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xD1ywMiX99s_00010_333_755.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_xD1ywMiX99s_00010_333_755", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/xD1ywMiX99s_00010_333_755", + "duration": 14.127 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xD1ywMiX99s_00010_333_755_joint0_skel.jpg", + "time": 8.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/xD1ywMiX99s_00010_333_755", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xJojrk13FG4_00030_0_148.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xJojrk13FG4_00030_0_148.json new file mode 100644 index 0000000000000000000000000000000000000000..9a327e11ea99ddcd36db5627c56d4bea2de3075c --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xJojrk13FG4_00030_0_148.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_xJojrk13FG4_00030_0_148", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/xJojrk13FG4_00030_0_148", + "duration": 5.039 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xJojrk13FG4_00030_0_148_joint0_skel.jpg", + "time": 3.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/xJojrk13FG4_00030_0_148", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xOfH2zAPyhw_00040_503_585.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xOfH2zAPyhw_00040_503_585.json new file mode 100644 index 0000000000000000000000000000000000000000..7ac74a65cac219fa37f2aa72796f88c27e3c2b4e --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xOfH2zAPyhw_00040_503_585.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_xOfH2zAPyhw_00040_503_585", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/xOfH2zAPyhw_00040_503_585", + "duration": 3.423 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xOfH2zAPyhw_00040_503_585_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/xOfH2zAPyhw_00040_503_585", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xPYSM_X_gnQ_00007_0_110.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xPYSM_X_gnQ_00007_0_110.json new file mode 100644 index 0000000000000000000000000000000000000000..d4e0399a7740f26b7f179817342c21cad0334fe1 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xPYSM_X_gnQ_00007_0_110.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_xPYSM_X_gnQ_00007_0_110", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/xPYSM_X_gnQ_00007_0_110", + "duration": 3.704 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is closing the refrigerator doors and walking away from the appliance.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xPYSM_X_gnQ_00007_0_110_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xPYSM_X_gnQ_00007_0_110_joint1_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/xPYSM_X_gnQ_00007_0_110", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xPtkAYDLzBQ_00005_310_395.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xPtkAYDLzBQ_00005_310_395.json new file mode 100644 index 0000000000000000000000000000000000000000..2c8d16e800ac8ce51f0fc22c7c425f2257368a45 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xPtkAYDLzBQ_00005_310_395.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_xPtkAYDLzBQ_00005_310_395", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/xPtkAYDLzBQ_00005_310_395", + "duration": 2.87 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xPtkAYDLzBQ_00005_310_395_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xPtkAYDLzBQ_00005_310_395_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/xPtkAYDLzBQ_00005_310_395", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xVH3jwjG8ok_00015_0_94.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xVH3jwjG8ok_00015_0_94.json new file mode 100644 index 0000000000000000000000000000000000000000..372985b7ba4451c261ec10d5da2bd3191d7c13e4 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xVH3jwjG8ok_00015_0_94.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_xVH3jwjG8ok_00015_0_94", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/xVH3jwjG8ok_00015_0_94", + "duration": 3.17 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xVH3jwjG8ok_00015_0_94_joint0_skel.jpg", + "time": 1.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xVH3jwjG8ok_00015_0_94_joint1_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/xVH3jwjG8ok_00015_0_94", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x_prt6sDXUQ_00059_174_390.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x_prt6sDXUQ_00059_174_390.json new file mode 100644 index 0000000000000000000000000000000000000000..64d9b5e292c21bf11fe60bec8e8e420c24cb3cd4 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_x_prt6sDXUQ_00059_174_390.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_x_prt6sDXUQ_00059_174_390", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/x_prt6sDXUQ_00059_174_390", + "duration": 8.64 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/x_prt6sDXUQ_00059_174_390_joint0_skel.jpg", + "time": 2.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/x_prt6sDXUQ_00059_174_390_joint1_skel.jpg", + "time": 4.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/x_prt6sDXUQ_00059_174_390_joint2_skel.jpg", + "time": 6.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/x_prt6sDXUQ_00059_174_390", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xfhzzAmk9GI_00021_559_843.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xfhzzAmk9GI_00021_559_843.json new file mode 100644 index 0000000000000000000000000000000000000000..c1cbfbca4bcf4e73ec94268d3a576f8653898844 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_xfhzzAmk9GI_00021_559_843.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_xfhzzAmk9GI_00021_559_843", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/xfhzzAmk9GI_00021_559_843", + "duration": 9.484 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xfhzzAmk9GI_00021_559_843_joint0_skel.jpg", + "time": 3.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xfhzzAmk9GI_00021_559_843_joint1_skel.jpg", + "time": 5.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/xfhzzAmk9GI_00021_559_843_joint2_skel.jpg", + "time": 7.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/xfhzzAmk9GI_00021_559_843", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zaJV54MKG-A_00012_0_149.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zaJV54MKG-A_00012_0_149.json new file mode 100644 index 0000000000000000000000000000000000000000..14d30400624680b751d27b43f9ae8851b8053bcd --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zaJV54MKG-A_00012_0_149.json @@ -0,0 +1,36 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_zaJV54MKG-A_00012_0_149", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/zaJV54MKG-A_00012_0_149", + "duration": 6.0 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zaJV54MKG-A_00012_0_149_joint0_skel.jpg", + "time": 1.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zaJV54MKG-A_00012_0_149_joint1_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/zaJV54MKG-A_00012_0_149", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zhvbtWYA1v4_00015_0_471.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zhvbtWYA1v4_00015_0_471.json new file mode 100644 index 0000000000000000000000000000000000000000..63b9dc926537d427b1982c12a5dc6dd56e517c0b --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zhvbtWYA1v4_00015_0_471.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_zhvbtWYA1v4_00015_0_471", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/zhvbtWYA1v4_00015_0_471", + "duration": 7.908 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zhvbtWYA1v4_00015_0_471_joint0_skel.jpg", + "time": 2.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/zhvbtWYA1v4_00015_0_471", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zkPa9-Y_M7k_00003_184_338.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zkPa9-Y_M7k_00003_184_338.json new file mode 100644 index 0000000000000000000000000000000000000000..70957a0ff45e358b561d438f33ac5e6a0f93dd8e --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zkPa9-Y_M7k_00003_184_338.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_zkPa9-Y_M7k_00003_184_338", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/zkPa9-Y_M7k_00003_184_338", + "duration": 3.1 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zkPa9-Y_M7k_00003_184_338_joint0_skel.jpg", + "time": 1.5 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/zkPa9-Y_M7k_00003_184_338", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zkPa9-Y_M7k_00064_0_146.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zkPa9-Y_M7k_00064_0_146.json new file mode 100644 index 0000000000000000000000000000000000000000..e4a08bd87191e149f149ae069586fcdc00f29ba6 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zkPa9-Y_M7k_00064_0_146.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_zkPa9-Y_M7k_00064_0_146", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/zkPa9-Y_M7k_00064_0_146", + "duration": 2.94 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zkPa9-Y_M7k_00064_0_146_joint0_skel.jpg", + "time": 0.8 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/zkPa9-Y_M7k_00064_0_146", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zoJXcEfo-Tg_00032_0_160.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zoJXcEfo-Tg_00032_0_160.json new file mode 100644 index 0000000000000000000000000000000000000000..6a26f8a95bccc6c0c6cc4e0748d2b10f705862f8 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zoJXcEfo-Tg_00032_0_160.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_zoJXcEfo-Tg_00032_0_160", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/zoJXcEfo-Tg_00032_0_160", + "duration": 5.341 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zoJXcEfo-Tg_00032_0_160_joint0_skel.jpg", + "time": 3.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/zoJXcEfo-Tg_00032_0_160", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zoplKrZlnQQ_00012_0_90.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zoplKrZlnQQ_00012_0_90.json new file mode 100644 index 0000000000000000000000000000000000000000..d61a01e20a446c9d7cca7d1261b08d53bfd3acb4 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zoplKrZlnQQ_00012_0_90.json @@ -0,0 +1,32 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_zoplKrZlnQQ_00012_0_90", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/zoplKrZlnQQ_00012_0_90", + "duration": 3.64 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "The person is covering a pot and moving it from the counter to the sink with a steady and focused demeanor.", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zoplKrZlnQQ_00012_0_90_joint0_skel.jpg", + "time": 2.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/zoplKrZlnQQ_00012_0_90", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zpjWgmz7_MQ_00026_0_119.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zpjWgmz7_MQ_00026_0_119.json new file mode 100644 index 0000000000000000000000000000000000000000..d86828af36e6d131430d92d145da37241fe3948c --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zpjWgmz7_MQ_00026_0_119.json @@ -0,0 +1,48 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_zpjWgmz7_MQ_00026_0_119", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/zpjWgmz7_MQ_00026_0_119", + "duration": 5.005 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zpjWgmz7_MQ_00026_0_119_joint0_skel.jpg", + "time": 0.4 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zpjWgmz7_MQ_00026_0_119_joint1_skel.jpg", + "time": 0.8 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zpjWgmz7_MQ_00026_0_119_joint2_skel.jpg", + "time": 1.2 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zpjWgmz7_MQ_00026_0_119_joint3_skel.jpg", + "time": 2.6 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zpjWgmz7_MQ_00026_0_119_joint4_skel.jpg", + "time": 3.4 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/zpjWgmz7_MQ_00026_0_119", + "rendered_video": "none" + } +} \ No newline at end of file diff --git a/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zta4suL4pxA_00001_454_1076.json b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zta4suL4pxA_00001_454_1076.json new file mode 100644 index 0000000000000000000000000000000000000000..595286f1f3fef648b3e14d4fb8cd7a9f295379b4 --- /dev/null +++ b/Tasks/Video/IMG_TXT_SKEL/L1_IMG_TXT_SKEL_zta4suL4pxA_00001_454_1076.json @@ -0,0 +1,40 @@ +{ + "metadata": { + "task_id": "L1_IMG_TXT_SKEL_zta4suL4pxA_00001_454_1076", + "task_level": "Level1", + "task_type": "IMG_TXT_SKEL", + "original_video": "none", + "original_motion": "Data_RAW/converted_motion/00000/zta4suL4pxA_00001_454_1076", + "duration": 10.402 + }, + "input": { + "prompts": { + "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with a text description of the overall full-body action, a target motion duration in seconds, and a sequence of keyframe images of a rendered skeleton extracted at specific timestamps. Your objective is to generate a continuous 3D motion sequence. You must ensure that the generated motion logically follows the text description, matches the target duration, matches the subject's poses in the keyframe images at their exact timestamps, and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the full-body motion.", + "modifier_instruction": "" + }, + "modalities": { + "text": "", + "audios": [], + "videos_processed": [], + "spatial_coordinates": [], + "input_images": [ + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zta4suL4pxA_00001_454_1076_joint0_skel.jpg", + "time": 2.5 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zta4suL4pxA_00001_454_1076_joint1_skel.jpg", + "time": 5.0 + }, + { + "path": "Data_Processed/Video/IMG_TXT_SKEL/zta4suL4pxA_00001_454_1076_joint2_skel.jpg", + "time": 8.0 + } + ] + } + }, + "ground_truth": { + "motion_parameters": "Data_RAW/converted_motion/00000/zta4suL4pxA_00001_454_1076", + "rendered_video": "none" + } +} \ No newline at end of file