MeshGraphormer Instructions for HandRefiner

December 29, 2023 ยท View on GitHub

Installation

Requirements

Install the MeshGraphormer to HandRefiner/MeshGraphormer

git clone --recursive https://github.com/microsoft/MeshGraphormer.git
cd MeshGraphormer
pip install ./manopth/.

Download

Make sure you are on 'HandRefiner/MeshGraphormer' directory for the following steps

  1. Create folder that store pretrained models.

    mkdir -p models  # pre-trained models

  2. Download pretrained models, and some code modifications.

    cp ../scripts/download_models.sh scripts/download_models.sh
cp ../scripts/_gcnn.py src/modeling/_gcnn.py
cp ../scripts/_mano.py src/modeling/_mano.py
cp ../scripts/config.py src/modeling/data/config.py
bash scripts/download_models.sh

    The resulting data structure should follow the hierarchy as below.

    MeshGraphormer 
|-- models  
|   |-- graphormer_release
|   |   |-- graphormer_hand_state_dict.bin
|   |-- hrnet
|   |   |-- hrnetv2_w64_imagenet_pretrained.pth
|   |   |-- cls_hrnet_w64_sgd_lr5e-2_wd1e-4_bs32_x100.yaml
|-- src 
|-- datasets 
|-- predictions 
|-- README.md 
|-- ... 
|-- ...

  3. Download MANO model from their official websites

    • Download MANO_RIGHT.pkl from MANO, and place it at MeshGraphormer/src/modeling/data.

    Please put the downloaded files under the MeshGraphormer/src/modeling/data directory. The data structure should follow the hierarchy below.

    MeshGraphormer  
|-- src  
|   |-- modeling
|   |   |-- data
|   |   |   |-- MANO_RIGHT.pkl
|-- models
|-- datasets
|-- predictions
|-- README.md 
|-- ... 
|-- ...

  4. exit the MeshGraphormer directory when finished

    cd ..