CFM

https://ieeexplore.ieee.org/document/10315169

---

This repository contains the official PyTorch implementation for CFM. Note that we currently only released the model and pre-trained model since it is easy to evaluate on your customized dataset. Please forgive the irregular variable naming in the code, it is too busy for me to re-organize the source code T.T.

Requirements

• timm == 0.5.4
• pytorch == 1.8.0

Pre-trained model preperation

The pretrained models can be access here https://drive.google.com/file/d/1iP8k4ffJ-3WcuEg5uyrsY3qOY0ALi83H/view?usp=sharing

Citation

If you find our work helpful in your research, please cite it as:

@ARTICLE{10315169,
  author={Luo, Anwei and Kong, Chenqi and Huang, Jiwu and Hu, Yongjian and Kang, Xiangui and Kot, Alex C.},
  journal={IEEE Transactions on Information Forensics and Security}, 
  title={Beyond the Prior Forgery Knowledge: Mining Critical Clues for General Face Forgery Detection}, 
  year={2024},
  volume={19},
  number={},
  pages={1168-1182},
  doi={10.1109/TIFS.2023.3332218}}