InvReg

Invariant Feature Regularization for Fair Face Recognition (ICCV'23)

Requirements

• numpy (1.21.4)
• easydict (1.9)
• onnx (1.10.1)
• scipy (1.6.3)
• tdqm (4.62.3)
• torch (1.11.0)
• mxnet (1.9.1)
• torchvision (0.11.0)
• sklearn (0.24.0)
• python (3.8.12)

Citation

If you find InvReg helpful in your research, please consider citing:

@inproceedings{ma2023invariant,
  title={Invariant Feature Regularization for Fair Face Recognition},
  author={Ma, Jiali and Yue, Zhongqi and Tomoyuki, Kagaya and Tomoki, Suzuki and Jayashree, Karlekar and Pranata, Sugiri and Zhang, Hanwang},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={20861--20870},
  year={2023}
}

References & Opensources

Part of our implementation is adopted from the Arcface, IP-IRM, TFace and InvariantRiskMinimization repositories.

• Arcface(https://github.com/deepinsight/insightface/tree/master/recognition/arcface_torch)
• IP-IRM(https://github.com/Wangt-CN/IP-IRM)
• TFace(https://github.com/Tencent/TFace/tree/v0.1.0)
• InvariantRiskMinimization(https://github.com/facebookresearch/InvariantRiskMinimization)

License

InvReg is .