Unsupervised Multimodal Deepfake Detection Through Explicit Intra-Modal and Cross-Modal Inconsistency Discovery


Mulin Tian (University of Southern California), Mahyar Khayatkhoei (University of Southern California), Joe Mathai (University of Southern California), Wael AbdAlmageed (Clemson University)
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)Supplementary

Abstract

Deepfake videos present an increasing threat to society with potentially negative impact on criminal justice, democracy, and personal safety and privacy. Meanwhile, detecting deepfakes, at scale, remains a very challenging task that often requires labeled training data from existing deepfake generation methods. Further, even the most accurate supervised deepfake detection methods do not generalize to deepfakes generated using new generation methods. In this paper, we propose a novel unsupervised method for detecting deepfake videos by explicitly discovering intra-modal and cross-modal inconsistencies between video segments. The fundamental hypothesis behind our proposed method is that motion or identity inconsistencies are inevitable in motion-transfer deepfake videos. We provide the first mathematical and empirical evidence to support this hypothesis, hence justifying the generalizability of our method. Our proposed method outperforms prior state-of-the-art unsupervised deepfake detection methods on the challenging FakeAVCeleb dataset, and also has several additional advantages: it is scalable because it does not require pristine (real) samples for each identity during inference and therefore can apply to arbitrarily many identities; generalizable because it is trained only on real videos and therefore does not rely on a particular deepfake method; reliable because it does not rely on any likelihood estimation in high dimensions; and explainable because it can pinpoint the exact location of modality inconsistencies.

Citation

@inproceedings{Tian_2025_BMVC,
author    = {Mulin Tian and Mahyar Khayatkhoei and Joe Mathai and Wael AbdAlmageed},
title     = {Unsupervised Multimodal Deepfake Detection Through Explicit Intra-Modal and Cross-Modal Inconsistency Discovery},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_722/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection