GEARS: Local geometry-aware Hand-Object Interaction Synthesis

Repo for "GEARS: Local geometry-aware Hand-Object Interaction Synthesis, CVPR'24"
[Paper] [Project Page]

Environment

This code is written and tested with Python 3.8. To install the required dependencies, run:

pip install -r requirements.txt

We additionally require the following libraries：

• PyTorch3D
• MPI-IS Mesh Processing Library
• Manopth layer for PyTorch

Please check the respective instructions for downloading and installation.

Data

1. Download the raw GRAB dataset and SMPL-X models by following instructions here.
2. Run our pre-processing code:

python data/create_dataset.py --grab_path $RAW_GRAB_FOLDER \
                                  --model_path $SMPLX_MODEL_FOLDER \
                                  --out_path $PROCESSED_GRAB_FOLDER

Training

Train the model with

python train.py --data_path $PROCESSED_GRAB_FOLDER

Citation

@inproceedings{zhou2024gears,
  title = {GEARS: Local Geometry-aware Hand-object Interaction Synthesis},
  author = {Zhou, Keyang and Bhatnagar, Bharat Lal and Lenssen, Jan Eric and Pons-Moll, Gerard},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2024},
}