文献类型：期刊文献

英文题名：FCL: frequency-based contrastive learning for generalizable face forgery detection

作者：Zhu, Yu[1];Wang, Shengze[1];Gu, Yufeng[1];Zhu, Ziming[1];Wang, Nan[1]

机构：[1]East China Univ Sci & Technol, Sch Elect & Commun Engn, Meilong Rd 130, Shanghai 200237, Peoples R China

年份：2026

卷号：32

期号：4

外文期刊名：MULTIMEDIA SYSTEMS

收录：;EI(收录号：20262020723615);WOS:【SCI-EXPANDED(收录号:WOS:001758540000016)】；

基金：This work was supported in part by the National Natural Science Foundation of China 62476088, and the Science and Technology Commission of Shanghai Municipality 20DZ2254400.

语种：英文

外文关键词：Face forgery detection; Frequency; Contrastive learning; Generalization ability

摘要：With the proliferation of deepfake techniques, face forgery detection has emerged as a critical area of research to mitigate the risks associated with facial manipulation. Recent studies have highlighted the superiority of frequency information over color-space input, especially in high-compression scenarios. However, most frequency-enhanced methods employ a two-stream network architecture, assigning separate encoders for RGB and frequency domains, and rely solely on vanilla binary cross-entropy loss, limiting their generalization ability. To address these limitations, we propose a novel framework, Frequency-enhanced Contrastive Learning (FCL), which trains the model in a supervised manner using contrastive loss. FCL treats two views of different modalities generated from the same image as positive pairs and samples with opposite labels as negative pairs. This approach pulls features from the RGB and high-frequency domains closer while pushing features of pristine and forgery faces apart. Additionally, we introduce a Shallow Feature Supplement (SFS) module to complement local information from low-level shallow feature embeddings into high-level feature maps, and a Dual Modal Fusion (DMF) module to adaptively aggregate information from both domains. Extensive experiments on seven datasets demonstrate that our method achieves competitive generalization performance against state-of-the-art competitors, with significant advantages in heavily compressed scenarios. FCL achieves an accuracy of 92.29% and an AUC of 94.25% on the heavily compressed FaceForensics++ (LQ, C40) dataset, which verifies its robustness against compression in practical application scenarios. The source code of our proposed algorithm is available at https://github.com/Y30230924Wang/Frequency-Enhanced-Contrastive-Learning.