Exposing the Deception: Uncovering More Forgery Clues for Deepfake Detection
Zhongjie Ba, Qingyu Liu, Zhenguang Liu, Shuang Wu, Feng Lin, Li Lu, Kui Ren
Abstract
Deepfake technology has given rise to a spectrum of novel and compelling applications. Unfortunately, the widespread proliferation of high-fidelity fake videos has led to pervasive confusion and deception, shattering our faith that seeing is believing. One aspect that has been overlooked so far is that current deepfake detection approaches may easily fall into the trap of overfitting, focusing only on forgery clues within one or a few local regions. Moreover, existing works heavily rely on neural networks to extract forgery features, lacking theoretical constraints guaranteeing that sufficient forgery clues are extracted and superfluous features are eliminated. These deficiencies culminate in unsatisfactory accuracy and limited generalizability in real-life scenarios.
In this paper, we try to tackle these challenges through three designs: (1) We present a novel framework to capture broader forgery clues by extracting multiple non-overlapping local representations and fusing them into a global semantic-rich feature. (2) Based on the information bottleneck theory, we derive Local Information Loss to guarantee the orthogonality of local representations while preserving comprehensive task-relevant information. (3) Further, to fuse the local representations and remove task-irrelevant information, we arrive at a Global Information Loss through the theoretical analysis of mutual information. Empirically, our method achieves state-of-the-art performance on five benchmark datasets. Our code is available at https://github.com/QingyuLiu/Exposing-the-Deception, hoping to inspire researchers.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
23 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Standing on the Shoulders of Giants: Reprogramming Visual-Language Model for General Deepfake Detection cites Detecting Deepfakes with Self-Blended Images
- Detecting Deepfakes with Self-Blended Images cites Learning Self-Consistency for Deepfake Detection
- Detecting Deepfakes with Self-Blended Images cites Leveraging Frequency Analysis for Deep Fake Image Recognition
- DeepShield: Fortifying Deepfake Video Detection with Local and Global Forgery Analysis cites Detecting Deepfakes with Self-Blended Images
- Exploring Unbiased Deepfake Detection via Token-Level Shuffling and Mixing cites Detecting Deepfakes with Self-Blended Images
- Exploring Unbiased Deepfake Detection via Token-Level Shuffling and Mixing cites Learning Self-Consistency for Deepfake Detection
- Exploring Unbiased Deepfake Detection via Token-Level Shuffling and Mixing cites End-to-End Reconstruction-Classification Learning for Face Forgery Detection
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes cites Detecting Deepfakes with Self-Blended Images
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes cites Learning Self-Consistency for Deepfake Detection
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes cites End-to-End Reconstruction-Classification Learning for Face Forgery Detection
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes cites Leveraging Frequency Analysis for Deep Fake Image Recognition
Cited by top-tier papers23
- Exploring Unbiased Deepfake Detection via Token-Level Shuffling and MixingXinghe Fu, Zhiyuan Yan, Taiping Yao, Shen Chen et al.AAAI 2025 · 41 citations
- Standing on the Shoulders of Giants: Reprogramming Visual-Language Model for General Deepfake DetectionKaiqing Lin, Yuzhen Lin, Weixiang Li, Taiping Yao et al.AAAI 2025 · 32 citations
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face DeepfakesLong Ma, Zhiyuan Yan, Jin Xu, Yize Chen et al.NeurIPS 2025 · 24 citations
- DeepShield: Fortifying Deepfake Video Detection with Local and Global Forgery AnalysisYinqi Cai, Jichang Li, Zhaolun Li, Weikai Chen et al.ICCV 2025 · 11 citations
- SpecXNet: A Dual-Domain Convolutional Network for Robust Deepfake DetectionInzamamul Alam, Md Tanvir Islam, Simon S. WooACM MM 2025 · 6 citations
Builds on32
- Leveraging Frequency Analysis for Deep Fake Image RecognitionJoel Frank, Thorsten Eisenhofer, Lea Schönherr, Asja Fischer et al.ICML 2020 · 848 citations
- FSGAN: Subject Agnostic Face Swapping and ReenactmentYuval Nirkin, Yosi Keller, Tal HassnerICCV 2019 · 710 citations
- Learning Self-Consistency for Deepfake DetectionTianchen Zhao, Xiang Xu, Mingze Xu, Hui Ding et al.ICCV 2021 · 368 citations
- Detecting Deepfakes with Self-Blended ImagesKaede Shiohara, Toshihiko YamasakiCVPR 2022 · 366 citations
- End-to-End Reconstruction-Classification Learning for Face Forgery DetectionJunyi Cao, Chao Ma, Taiping Yao, Shen Chen et al.CVPR 2022 · 327 citations
Related papers
- UCF: Uncovering Common Features for Generalizable Deepfake DetectionZhiyuan Yan, Yong Zhang, Yanbo Fan, Baoyuan WuICCV 2023 · 264 citations
- Multi-Attentional Deepfake DetectionHanqing Zhao, Wenbo Zhou, Dongdong Chen, Tianyi Wei et al.CVPR 2021
- SUMI-IFL: An Information-Theoretic Framework for Image Forgery Localization with Sufficiency and Minimality ConstraintsZiqi Sheng, Wei Lu, Xiangyang Luo, Jiantao Zhou et al.AAAI 2025 · 3 citations
- Locate and Verify: A Two-Stream Network for Improved Deepfake DetectionChao Shuai, Jieming Zhong, Shuang Wu, Feng Lin et al.ACM MM 2023 · 52 citations
- Spatio-Temporal Catcher: A Self-Supervised Transformer for Deepfake Video DetectionMaosen Li, Xurong Li, Kun Yu, Cheng Deng et al.ACM MM 2023 · 9 citations
