Dataset Card for VR-Folding Dataset

Dataset Summary

This is the VR-Folding dataset created by the CVPR 2023 paper GarmentTracking: Category-Level Garment Pose Tracking. This dataset is recorded with a system called VR-Garment, which is a garment-hand interaction environment based on Unity.

To download the dataset, use the following shell snippet:

git lfs install
git clone https://huggingface.co/datasets/robotflow/garment-tracking
# if you want to clone without large files – just their pointers
# prepend your git clone with the following env var: GIT_LFS_SKIP_SMUDGE=1

# merge multiple .zip files (e.g. folding) into one .zip file
cd data/folding
cat folding_dataset.z*  > folding_dataset.zip
# unzip
unzip folding_dataset.zip

All the data are stored in zarr format.

Dataset Structure

Here is the detailed stucture of a data example (zarr format) of one frame:

00068_Tshirt_000000_000000
 ├── grip_vertex_id
 │   ├── left_grip_vertex_id (1,) int32
 │   └── right_grip_vertex_id (1,) int32
 ├── hand_pose
 │   ├── left_hand_euler (25, 3) float32
 │   ├── left_hand_pos (25, 3) float32
 │   ├── right_hand_euler (25, 3) float32
 │   └── right_hand_pos (25, 3) float32
 ├── marching_cube_mesh
 │   ├── is_vertex_on_surface (6410,) bool
 │   ├── marching_cube_faces (12816, 3) int32
 │   └── marching_cube_verts (6410, 3) float32
 ├── mesh
 │   ├── cloth_faces_tri (8312, 3) int32
 │   ├── cloth_nocs_verts (4434, 3) float32
 │   └── cloth_verts (4434, 3) float32
 └── point_cloud
     ├── cls (30000,) uint8
     ├── nocs (30000, 3) float16
     ├── point (30000, 3) float16
     ├── rgb (30000, 3) uint8
     └── sizes (4,) int64

Specifically, we render 4-view RGB-D images with Unity and generate concated point clouds for each frame. Here grip_vertex_id is the vertex index list of the grasped points of the mesh.

Dataset Example

Please see example for example data and visualization scripts. Here are two video examples for flattening and folding task.