FaceRSA: RSA-Aware Facial Identity Cryptography Framework (AAAI 24)

Getting Started

Prerequisites

$ conda install --yes -c pytorch pytorch=1.7.1 torchvision cudatoolkit=11.0
$ pip install pyyaml

Pretrained Model

Please download the pre-trained model from the following link, and place it under ./pretrained_models/.

Path	Description
FaceRSA	Our pre-trained FaceRSA model.

Auxiliary Models and Latent Codes

In addition, we provide various auxiliary models and latent codes inverted by e4e.

Path	Description
FFHQ StyleGAN	StyleGAN model pretrained on FFHQ taken from rosinality with 1024x1024 output resolution.

Testing

Inference

The results are saved in ./sample/anonymized/ and ./sample/recovered/

Example of Using FaceRSA with random password

python inference.py

Citation

@inproceedings{zhang2024facersa,
  title={FaceRSA: RSA-Aware Facial Identity Cryptography Framework},
  author={Zhang, Zhongyi and Wei, Tianyi and Zhou, Wenbo and Zhao, Hanqing and Zhang, Weiming and Yu, Nenghai},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={38},
  number={7},
  pages={7423--7431},
  year={2024}
}

Training code will be released soon.