Disentangled Mesh Variational Autoencoder

Official Pytorch implementation of the paper "Disentangled representations: towards interpretation of sex determination from hip bone"

Bibtex

If you find this code useful in your research, please cite:

@article{zou2023disentangled,
  title={Disentangled representations: towards interpretation of sex determination from hip bone},
  author={Zou, Kaifeng and Faisan, Sylvain and Heitz, Fabrice and Epain, Marie and Croisille, Pierre and Fanton, Laurent and Valette, S{\'e}bastien},
  journal={The Visual Computer},
  pages={1--15},
  year={2023},
  publisher={Springer}
}

1. Requirements

This code is tested on Python3.8, Pytorch versoin 1.11.0+cu113, torch-geometric version 2.0.4 . Requirments can be install by running

  pip install -r requirements.txt

Install mesh processing libraries from MPI-IS/mesh. Note that the python3 version of mesh package library is needed for this.

2. Download the datasets

Since we test on a private dataset, the original dataset is not available. However we provide a fake dataset generated by our Mesh VAE to validate the algorithm. You can download it from google drive

3. Training

python main.py -- train

4. Testing

python main.py -- test --vis

Note that the visualization functions only when the test mode is enabled.

5. Inference

 python inference.py --error_list --inference --data_dir ./data/batch3 --output_path ./

valette/Mesh-VAE