Official Implementation of DFREC:DeepFake Identity Recovery Based on Identity-aware Masked Autoencoder
Recent advances in deepfake forensics have primarily focused on improving the classification accuracy and generalization performance. Despite enormous progress in detection accuracy across a wide variety of forgery algorithms, existing algorithms lack intuitive interpretability and identity traceability to help with forensic investigation. In this work, we introduce a novel DeepFake Identity Recovery scheme (DFREC) to fill this gap. DFREC aims to recover the pair of source and target faces from a deepfake image to facilitate deepfake identity tracing and reduce the risk of deepfake attack. We evaluate DFREC on six different high-fidelity face-swapping attacks on FaceForensics++, CelebaMegaFS and FFHQ-E4S datasets, which demonstrate its superior recovery performance over state-of-the-art deepfake recovery algorithms.
checkpoint path：https://drive.google.com/drive/folders/1ZXH-7QTy5P-o1zfhY7myUeTIhXUj_JmS?usp=sharing
To inference the DFREC, run this command:
#unzip the segmentation models codes
unzip segmentation_models.zip
#Prepare the environment required for the project
python -m pip install -r requirements.txt
#and then you can run the inference code
python DFREC_eval.pyThe training code will be released after the paper is accepted.
@article{yu2024dfrec,
title={DFREC: DeepFake Identity Recovery Based on Identity-aware Masked Autoencoder},
author={Yu, Peipeng and Gao, Hui and Huang, Zhitao and Xia, Zhihua and Chang, Chip-Hong},
journal={arXiv preprint arXiv:2412.07260},
year={2024}
}