DCHVF-GAN: Synthesizing Adversarial DeepFakes with High Visual Fidelity by Multimodality Fusion
ACM Transactions on Multimedia Computing, Communications, and Applications · 30.06.2026 · 22 (6) · 1-19
DOI: 10.1145/3808697 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM Transactions on Multimedia Computing, Communications, and Applications, online erschienen am 21.05.2026. Verfasst von Feng Ding und 6 weiteren Autor:innen. Die Arbeit nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Jiangxi Provincial Natural Science Foundation.
Abstract (Originalsprache)
DeepFake, an AI-driven face-swapping technique, has been weaponized to spread disinformation. In response, researchers have developed forensic detectors to identify such manipulations. To circumvent these defenses, a growing body of work now focuses on generating adversarial samples—carefully perturbed forgeries designed to deceive detection tools. However, most existing adversarial generation methods sacrifice image quality to achieve undetectability, introducing perceptible artifacts that ironically make them more detectable under human scrutiny. To address this limitation, we propose a novel spectral fusion approach to multimodally synthesize forgery traces from authentic facial images. Unlike traditional noise injection methods, our technique integrates diffusion-based noise during image preprocessing, embedding perturbations in the forward process of a diffusion model. This approach not only deceives forensic detectors more effectively but also preserves high visual fidelity. Through extensive experiments, our method achieves state-of-the-art DeepFake anti-forensic performance while preserving high visual fidelity, ensuring that the adversarial samples remain indistinguishable from real images.
Förderung
- National Natural Science Foundation of China
- Jiangxi Provincial Natural Science Foundation
Zitieren
APA
Ding, F., He, X., Kuang, R., Xiao, M., Zhu, X., Zhu, G., & Atrey, P. K. (2026). DCHVF-GAN: Synthesizing Adversarial DeepFakes with High Visual Fidelity by Multimodality Fusion. ACM Transactions on Multimedia Computing, Communications, and Applications, 22(6), 1-19. https://doi.org/10.1145/3808697
Vancouver
Ding F, He X, Kuang R, Xiao M, Zhu X, Zhu G, et al. DCHVF-GAN: Synthesizing Adversarial DeepFakes with High Visual Fidelity by Multimodality Fusion. ACM Transactions on Multimedia Computing, Communications, and Applications. 2026;22(6):1-19. doi:10.1145/3808697
BibTeX
@article{Ding2026,
author = {Ding, Feng and He, Xinan and Kuang, Rensheng and Xiao, Mengyao and Zhu, Xiaogang and Zhu, Guopu and Atrey, Pradeep K.},
title = {DCHVF-GAN: Synthesizing Adversarial DeepFakes with High Visual Fidelity by Multimodality Fusion},
journal = {ACM Transactions on Multimedia Computing, Communications, and Applications},
year = {2026},
volume = {22},
number = {6},
pages = {1--19},
doi = {10.1145/3808697}
}
EUVIMED