This repository holds the PyTorch implementation of our paper Hierarchical Multi-Task Network For Race, Gender and Facial Attractiveness Recognition.

1. By leveraging FiveCrops inference, we are able to achieve better performance.
2. We also report 5 cross validation results, since we find newly proposed models often use this metric instead of 6/4 splitting strategy.

• Install 3rd party libraries
  sudo pip3 install -r requirements.txt
• Modify cfg.py to fit your path

• Pretrained HMTNet

If you find this repository helps your research, please cite our paper:

@inproceedings{xu2019hierarchical,
  title={Hierarchical Multi-Task Network For Race, Gender and Facial Attractiveness Recognition},
  author={Xu, Lu and Fan, Heng and Xiang, Jinhai},
  booktitle={2019 IEEE International Conference on Image Processing (ICIP)},
  pages={3861--3865},
  year={2019},
  organization={IEEE}
}