TOHGS

Task-Oriented Human Grasp Synthesis via Context- and Task-Aware Diffusers

Env info

• Ubuntu 20.04
• Pytorch 2.0.1
• CUDA 11.7
• Python 3.8

# create conda env
conda create -n tohgs python=3.8
conda activate tohgs

Install

1. MANO

cd manopth
pip install .

2. Install Pytorch, Pytorch3D and Kaolin

pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2
pip install fvcore
pip install --no-index --no-cache-dir pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu117_pyt201/download.html
pip install kaolin==0.15.0 -f https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-2.0.1_cu117.html

3. Other Python packages

pip install -r requirements.txt

4. PointNet++

git clone https://github.com/erikwijmans/Pointnet2_PyTorch.git
cd Pointnet2_PyTorch
pip install -e .
pip install pointnet2_ops_lib/.

5. V-hacd

git clone https://github.com/kmammou/v-hacd.git
cd v-hacd
cd app
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build

Download require files

1. Download MANO model from Link unzip and copy the file MANO_RIGHT.pkl into TOHGS/models/mano
2. Download object meshes file from Link unzip and copy all of folders into TOHGS/models/
3. Download dataset from Link unzip and copy folder task_oriented_grasps_dataset into TOHGS/