KailinLi / mano2nimble

Transfer the MANO hand poses to the NIMBLE hand poses

mano2nimble

This repository is a submodule of the CHORD project.

This repository contains the code for the conversion of the MANO model to the Nimble model. The conversion is done using the Nimble library and the MANO model.

Prerequisites

• Python 3
• pytorch
• numpy
• manotorch
• pytorch3d
• open3d
• tqdm
• trimesh

Downloads

• Download the MANO model from the MANO website.
• Download the Nimble model from the Nimble library

Your directory should look like this:

.
├── assets
│   └── mano_v1_2
│       ├── models
│           └── MANO_RIGHT.pkl
├── nimble
│   ├── assets
│   │   ├── NIMBLE_DICT_9137.pkl
│   │   ├── ...

Usage

Simply run the following command:

python mano2nimble.py -v

Aknowledgements

This code is based on the manotorch and Nimble library.

If you find this code useful, consider citing the following related works:

@inproceedings{li2023chord,
  title={Chord: Category-level hand-held object reconstruction via shape deformation},
  author={Li, Kailin and Yang, Lixin and Zhen, Haoyu and Lin, Zenan and Zhan, Xinyu and Zhong, Licheng and Xu, Jian and Wu, Kejian and Lu, Cewu},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={9444--9454},
  year={2023}
}
@article{li2024semgrasp,
  title={SemGrasp: Semantic Grasp Generation via Language Aligned Discretization},
  author={Li, Kailin and Wang, Jingbo and Yang, Lixin and Lu, Cewu and Dai, Bo},
  journal={arXiv preprint arXiv:2404.03590},
  year={2024}
}