mikecheninoulu / CGT

Code for Contrastive-Geometry Networks for Generalized 3D Pose Transfer

CGTransformer

Code for our AAAI 2022 paper "Geometry-Contrastive Transformer network for Generalized 3D Pose Transfer"

Contrastive-Geometry Transformer

This is the PyTorch implementation of our AAAI 2022 paper Geometry-Contrastive Transformer for Generalized 3D Pose Transfer. Haoyu Chen, Hao Tang, Zitong Yu, Nicu Sebe, Guoying Zhao.

Citation

If you use our code or paper, please consider citing:

@inproceedings{chen2021GCN,
  title={Geometry-Contrastive Transformer for Generalized 3D Pose Transfer},
  author={Chen, Haoyu and Tang, Hao and Yu, Zitong and Sebe, Nicu and Zhao, Guoying},
  booktitle={AAAI},
  year={2022}
}

Dependencies

Requirements:

• python3.6
• numpy
• pytorch==1.1.0 and above
• trimesh

Dataset preparation

We use the SMPL-NPT dataset provided by NPT, please download data from this link http://www.sdspeople.fudan.edu.cn/fuyanwei/download/NeuralPoseTransfer/data/,

Training

The usage of our code is easy, just run the code below.

python train.py

Evaluation

We use the same evaluation protocol as NPT for both seen and unseen settings.

Run the code below to conduct the evaluation.

python evaluation_NPT.py

Acknowledgement

Part of our code is based on

3D transfer: NPT，

Transformer framework: (https://github.com/lucidrains/vit-pytorch)

Many thanks!

License

MIT-2.0 License