Only Train Once: Uncertainty-Aware One-Class Learning for Face Authenticity Detection

2026-05-11

AI Summary

This work addresses the limitations of existing face forgery detection methods, which rely on supervised binary classification and exhibit poor generalization to unseen generation paradigms while struggling to handle diverse forgery types in a unified manner. The authors reformulate the task as a one-class classification problem requiring only genuine face images and propose a self-supervised framework, FADNet, that uniquely integrates one-class learning with evidential deep learning (EDL). A plug-and-play pseudo-forgery image generator (PFIG) is introduced to sharpen the decision boundary without using any forged samples during training. Evaluated on the DF40 and ASFD benchmarks, the method achieves average accuracies of 96.63% and average precisions of 98.83%, respectively, significantly outperforming current state-of-the-art approaches.

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\%.

Problem

• face forgery detection
• generalization
• one-class classification
• unseen generative paradigms
• identity fraud

Innovation

• One-Class Classification
• Evidential Deep Learning
• Face Forgery Detection
• Uncertainty Quantification
• Pseudo-Forgery Generation