Rethinking Domain Generalization for Face Anti-spoofing: Separability and Alignment

This is the source code for CVPR 2023 paper Rethinking Domain Generalization for Face Anti-spoofing: Separability and Alignment by Yiyou Sun, Yaojie Liu, Xiaoming Liu, Yixuan Li and Wen-Sheng Chu.

Preliminaries

It is tested under Ubuntu Linux 20.04 and Python 3.8 environment, and requries some packages to be installed:

• PyTorch
• scipy
• numpy
• sklearn
• MTCNN
• ylib (Manually download and copy to the current folder)

Usage

1. Dataset Preparation

Download the OULU-NPU, CASIA-FASD, Idiap Replay-Attack, and MSU-MFSD datasets. Put datasets into the directory of datasets/FAS.

2. Prepocessing

Run ./preposess.py.

3. Demo

Run ./train.py --protocol [O_C_I_to_M/O_M_I_to_C/O_C_M_to_I/I_C_M_to_O].

Citation

If you use our codebase, please cite our work:

@article{sun2023safas,
  title={Rethinking Domain Generalization for Face Anti-spoofing:
Separability and Alignment},
  author={Sun, Yiyou and Liu, Yaojie and Liu, Xiaoming and Li, Yixuan and Chu Wen-Sheng},
  journal={CVPR},
  year={2023}
}