| Humans with Collisions (HwC) Pose & Motion Dataset for PoseShield | |
| Website: https://huggingface.co/datasets/ZYYY99/Humans_with_Collision | |
| This repository contains the training, evaluation, and benchmark dataset for "PoseShield: Neural Collision Fields for Human Self-Collision Resolution (ECCV 2026)". | |
| Dataset Structure: | |
| - data/dataset/train_list.csv - Train set list | |
| - data/dataset/test_list.csv - Test set list | |
| - data/dataset/augmented_data/ - Self-colliding poses (.npz) used for training | |
| - data/dataset/gt_data/ - Corresponding collision-free ground truth poses (.npz) | |
| - data/dataset_test/ - HwC 500-pose benchmark subset (.pkl, .obj, .png) | |
| - data/motion_canonical/ - 100 canonical MotionFix motion sequences (.npy) | |
| Usage Instructions: | |
| To use this dataset in your PoseShield project, clone this Hugging Face repository directly or download the files and place them under your local `data/` directory: | |
| git lfs install | |
| git clone https://huggingface.co/datasets/ZYYY99/Humans_with_Collision | |
| Alternatively, you can download the dataset programmatically using the huggingface_hub library: | |
| from huggingface_hub import snapshot_download | |
| snapshot_download( | |
| repo_id="ZYYY99/Humans_with_Collision", | |
| repo_type="dataset", | |
| local_dir="data" | |
| ) | |