Only Train Once: Uncertainty-Aware One-Class Learning for Face Authenticity Detection
2026-05-11 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed FADNet, a new method to detect fake or forged faces in images by only learning from real faces, instead of trying to spot every possible type of fake. Their approach treats the problem as identifying anything that looks very different from real faces, which helps it catch fakes from unknown generation methods. They also use a technique to measure how uncertain the model is about its decisions and a tool to create fake-like images to improve detection. Tests show FADNet works better and more reliably than previous models on big face forgery datasets.
face forgery detectionone-class classificationgenerative paradigmsDeepFakesEvidential Deep Learningpseudo-forgery image generatorfeature embeddingidentity fraudbinary classificationgeneralization
Authors
Qingchao Jiang, Zhenxuan Hou, Zhiying Zhu, Zhenxing Qian, Xinpeng Zhang, Zaiwang Gu
Abstract
The rapid evolution of generative paradigms has enabled the creation of highly realistic imagery, which escalating the risks of identity fraud and the dissemination of disinformation. Most existing approaches frame face forgery detection as a fully supervised binary classification problem. Consequently, these models typically exhibit significant performance decay when tasked with detecting forgeries from previously unseen generative paradigms. Furthermore, these methods focus exclusively on either DeepFakes or fully synthesized faces, thereby failing to provide a generalized framework for universal face forgery detection. In this paper, we address this challenge by introducing FADNet (Face Authenticity Detector Net), % a self-supervised framework that which reformulates face forgery detection as a one-class classification (OCC) task. By training exclusively on authentic facial data to capture their intrinsic representations, FADNet flags any image whose feature embedding deviates significantly from the learned distribution of real faces as a forgery. The framework incorporates Evidential Deep Learning (EDL) to quantify predictive uncertainty and utilizes a plug-and-play pseudo-forgery image generator (PFIG) to tighten decision boundaries around authentic data. Extensive experimental evaluations on the DF40 and ASFD benchmarks demonstrate that FADNet achieves superior performance and generalization capabilities. Specifically, FADNet substantially outperforms existing state-of-the-art (SOTA) methods, yielding a remarkable average accuracy of 96.63\% and an average precision of 98.83\%.