{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,4]],"date-time":"2026-06-04T16:00:29Z","timestamp":1780588829736,"version":"3.54.1"},"reference-count":39,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,12,10]],"date-time":"2025-12-10T00:00:00Z","timestamp":1765324800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Big Data"],"abstract":"<jats:p>Deepfake detection systems have achieved impressive accuracy on conventional forged images; however, they remain vulnerable to anti-forensic or adversarial samples deliberately crafted to evade detection. Such samples introduce imperceptible perturbations that conceal forgery artifacts, causing traditional binary classifiers\u2014trained solely on real and forged data\u2014to misclassify them as authentic. In this paper, we address this challenge by proposing a multi-channel feature extraction framework combined with a three-class classification strategy. Specifically, one channel focuses on extracting identity-preserving facial representations to capture inconsistencies in personal identity traits, while additional channels extract complementary spatial and frequency domain features to detect subtle forgery traces. These multi-channel features are fused and fed into a three-class detector capable of distinguishing real, forged, and anti-forensic samples. Experimental results on datasets incorporating adversarial deepfakes demonstrate that our method substantially improves robustness against anti-forensic attacks while maintaining high accuracy on conventional deepfake detection tasks.<\/jats:p>","DOI":"10.3389\/fdata.2025.1720525","type":"journal-article","created":{"date-parts":[[2025,12,10]],"date-time":"2025-12-10T05:26:09Z","timestamp":1765344369000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Detecting anti-forensic deepfakes with identity-aware multi-branch networks"],"prefix":"10.3389","volume":"8","author":[{"given":"Mingyu","family":"Zhu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jun","family":"Long","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2025,12,10]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","DOI":"10.48850\/arXiv.2107.02045","article-title":"Understanding the security of deepfake detection","author":"Cao","year":"2021","journal-title":"arXiv [preprint]"},{"key":"B2","doi-asserted-by":"crossref","first-page":"39","DOI":"10.1109\/SP.2017.49","article-title":"\u201cTowards evaluating the robustness of neural networks,\u201d","volume-title":"2017 IEEE Symposium on Security and Privacy (SP)","author":"Carlini","year":"2017"},{"key":"B3","doi-asserted-by":"publisher","first-page":"103097","DOI":"10.1016\/j.jvcir.2021.103097","article-title":"Hybrid prediction-based pixel-value-ordering method for reversible data hiding","volume":"77","author":"Chang","year":"2021","journal-title":"J. Vis. Commun. Image Represent"},{"key":"B4","unstructured":"2017"},{"key":"B5","first-page":"4690","article-title":"\u201cArcface: additive angular margin loss for deep face recognition,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Deng","year":"2019"},{"key":"B6","doi-asserted-by":"publisher","first-page":"6682","DOI":"10.1109\/TII.2022.3201572","article-title":"Securing facial bioinformation by eliminating adversarial perturbations","volume":"19","author":"Ding","year":"","journal-title":"IEEE Trans. Ind. Inform"},{"key":"B7","doi-asserted-by":"publisher","first-page":"2485","DOI":"10.1109\/LSP.2024.3438561","article-title":"Disrupting anti-spoofing systems by images of consistent identity","volume":"31","author":"Ding","year":"2024","journal-title":"IEEE Signal Process. Lett"},{"key":"B8","doi-asserted-by":"publisher","first-page":"7162","DOI":"10.1109\/TCYB.2022.3210294","article-title":"ExS-GAN: synthesizing anti-forensics images via extra supervised GAN","volume":"53","author":"Ding","year":"","journal-title":"IEEE Trans. Cybern"},{"key":"B9","doi-asserted-by":"publisher","first-page":"4715","DOI":"10.1109\/TCSVT.2019.2963715","article-title":"Meteor: measurable energy map toward the estimation of resampling rate via a convolutional neural network","volume":"30","author":"Ding","year":"2020","journal-title":"IEEE Trans. Circuits Syst. Video Technol"},{"key":"B10","doi-asserted-by":"publisher","first-page":"3429","DOI":"10.1109\/TMM.2021.3098422","article-title":"Anti-forensics for face swapping videos via adversarial training","volume":"24","author":"Ding","year":"2021","journal-title":"IEEE Trans. Multimedia"},{"key":"B11","first-page":"7890","article-title":"\u201cWatch your up-convolution: CNN based generative deep neural networks are failing to reproduce spectral distributions,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Durall","year":"2020"},{"key":"B12","doi-asserted-by":"crossref","DOI":"10.1145\/3729233","article-title":"\u201cGenerating higher-quality anti-forensics deepfakes with adversarial sharpening mask,\u201d","volume-title":"ACM Transactions on Multimedia Computing, Communications and Applications","author":"Fan","year":"2025"},{"key":"B13","first-page":"4545","article-title":"\u201cSynthesizing black-box anti-forensics deepfakes with high visual quality,\u201d","volume-title":"ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","author":"Fan","year":"2024"},{"key":"B14","doi-asserted-by":"crossref","first-page":"114","DOI":"10.1109\/MIPR59079.2023.00035","article-title":"\u201cAttacking identity semantics in deepfakes via deep feature fusion,\u201d","volume-title":"2023 IEEE 6th International Conference on Multimedia Information Processing and Retrieval (MIPR)","author":"Fan","year":"2023"},{"key":"B15","first-page":"10196","article-title":"\u201cHierarchical representations with statistical modeling for face forgery detection,\u201d","author":"Guo","year":"2021","journal-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition"},{"key":"B16","first-page":"5039","article-title":"\u201cLips don't lie: a generalisable and forgery-agnostic face forgery detection method,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Haliassos","year":"2021"},{"key":"B17","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2503.06142","article-title":"Vlforgery face triad: detection, localization and attribution via multimodal large language models","author":"He","year":"2025","journal-title":"arXiv [preprint]"},{"key":"B18","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2304.11670","article-title":"Evading deepfake detectors via adversarial statistical consistency","author":"Hou","year":"2023","journal-title":"arXiv [preprint]"},{"key":"B19","first-page":"4490","article-title":"\u201cImplicit identity driven deepfake face swapping detection,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Huang","year":"2023"},{"key":"B20","article-title":"\u201cRobust deepfake detection against adversarial attacks via multi-modal learning,\u201d","author":"Huang","year":"2024","journal-title":"IEEE Transactions on Information Forensics and Security"},{"key":"B21","first-page":"4401","article-title":"\u201cA style-based generator architecture for generative adversarial networks,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Karras","year":"2019"},{"key":"B22","doi-asserted-by":"crossref","first-page":"1006","DOI":"10.1145\/3652583.3658035","article-title":"\u201cClipping the deception: adapting vision-language models for universal deepfake detection,\u201d","volume-title":"Proceedings of the 2024 International Conference on Multimedia Retrieval","author":"Khan","year":"2024"},{"key":"B23","article-title":"\u201cExposing deepfake videos by detecting face warping artifacts,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops","author":"Li","year":"2018"},{"key":"B24","first-page":"3207","article-title":"\u201cCeleb-df: a large-scale challenging dataset for deepfake forensics,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Li","year":"2020"},{"key":"B25","doi-asserted-by":"publisher","DOI":"10.4855\/arXiv.2402.00045","article-title":"Detecting multimedia generated by large ai models: a survey","author":"Lin","year":"","journal-title":"arXiv [preprint]"},{"key":"B26","first-page":"16815","article-title":"\u201cPreserving fairness generalization in deepfake detection,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Lin","year":""},{"key":"B27","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2406.06965","article-title":"Evolving from single-modal to multi-modal facial deepfake detection: a survey","author":"Liu","year":"","journal-title":"arXiv [preprint]"},{"key":"B28","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2406.13295","article-title":"Media forensics and deepfake systematic survey. archive of deepfake detection","author":"Liu","year":"","journal-title":"arXiv [preprint]"},{"key":"B29","first-page":"8687","article-title":"\u201cThinking in frequency: face forgery detection by mining frequency-aware clues,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Qian","year":"2020"},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.4850\/arXiv.2204.06125","article-title":"Hierarchical text-conditional image generation with clip latents","author":"Ramesh","year":"2022","journal-title":"arXiv [preprint]"},{"key":"B31","first-page":"10684","article-title":"\u201cHigh-resolution image synthesis with latent diffusion models,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Rombach","year":"2022"},{"key":"B32","first-page":"1","article-title":"\u201cFaceforensics++: learning to detect manipulated facial images,\u201d","volume-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV)","author":"R\u00f6ssler","year":"2019"},{"key":"B33","doi-asserted-by":"publisher","DOI":"10.1016\/j.knosys.2025.113922","article-title":"Diffusion model in modern detection: Advancing deepfake techniques","author":"Siddiqui","year":"","journal-title":"Knowl-Based Syst"},{"key":"B34","doi-asserted-by":"publisher","first-page":"126150","DOI":"10.1016\/j.eswa.2024.126150","article-title":"Enhanced deepfake detection with densenet and cross-vit","volume":"267","author":"Siddiqui","year":"","journal-title":"Expert Syst. Appl"},{"key":"B35","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1312.6199","article-title":"Intriguing properties of neural networks","author":"Szegedy","year":"2014","journal-title":"arXiv [preprint]"},{"key":"B36","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv:2206.00477","article-title":"Anti-forgery: towards a stealthy and robust deepfake disruption attack via adversarial perceptual-aware perturbations","author":"Wang","year":"2022","journal-title":"arXiv [preprint]"},{"key":"B37","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2406.09739","article-title":"Decoupling forgery semantics for generalizable deepfake detection","author":"Ye","year":"2024","journal-title":"arXiv [preprint]"},{"key":"B38","doi-asserted-by":"crossref","first-page":"169","DOI":"10.1145\/3577163.3595103","article-title":"\u201cExposing deepfakes using dual-channel network with multi-axis attention and frequency analysis,\u201d","volume-title":"Proceedings of the 2023 ACM Workshop on Information Hiding and Multimedia Security","author":"Zhou","year":"2023"},{"key":"B39","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv:2506.00874","article-title":"Breaking latent prior bias in detectors for generalizable aigc image detection","author":"Zhou","year":"2025","journal-title":"arXiv [preprint]"}],"container-title":["Frontiers in Big Data"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2025.1720525\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,10]],"date-time":"2025-12-10T05:26:18Z","timestamp":1765344378000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2025.1720525\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,10]]},"references-count":39,"alternative-id":["10.3389\/fdata.2025.1720525"],"URL":"https:\/\/doi.org\/10.3389\/fdata.2025.1720525","relation":{},"ISSN":["2624-909X"],"issn-type":[{"value":"2624-909X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,10]]},"article-number":"1720525"}}