Article: Diffusion-generated face image identification technique based on diffusion reconstruction error feature Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.14 pp.20 - 43 Abstract: Traditional deep learning detectors often struggle to generalise when detecting diffusion-generated content. To address this, we propose DIRE, a generalised detector leveraging reconstruction error image representation. The framework standardises facial feature spaces through constrained feature learning and introduces a gradient suppression algorithm to filter abnormal gradients, preventing shortcut learning and enhancing generalisable feature extraction. Experiments on hybrid datasets validate DIRE's effectiveness in four cross-domain tasks (O&C&M→I, O&C&I→M, O&I&M→C, and I&C&M→O). Ablation studies confirm the synergy of feature standardisation and gradient suppression, reducing bias by 97.6% and parameters by 42%, while accelerating inference by 2.3×. DIRE achieves 98.2% and 96.7% accuracy on two tasks (O&C&I→M and O&M&I→C), outperforming state-of-the-art methods by 5.3% while maintaining computational efficiency. This study advances generative face detection through dual optimisation, offering a lightweight framework for financial identity verification and social media content moderation. Inderscience Publishers - linking academia, business and industry through research

Title: Diffusion-generated face image identification technique based on diffusion reconstruction error feature

Authors: Yunyue Peng; Yuan Liu; Tianqi Xie; Wei Zeng

Addresses: College of Computer Science and Cyber (Pilot Software College), Computer Science and Technology, Chengdu University of Technology, Chengdu, Sichuan, 610059, China ' School of Computer Science and Engineering (School of Cyberspace Security), University of Electronic Science and Technology of China, Chengdu, Sichuan, 610059, China ' Information & Communication Department, China Electric Power Research Institute, Beijing, 100192, China ' College of Computer Science and Cyber Security (Pilot Software College), Chengdu University of Technology, Chengdu, Sichuan, 610059, China; Sichuan Engineering Technology Research Center of Industrial Internet Intelligent Monitoring and Application, Chengdu, Sichuan, 610059, China

Abstract: Traditional deep learning detectors often struggle to generalise when detecting diffusion-generated content. To address this, we propose DIRE, a generalised detector leveraging reconstruction error image representation. The framework standardises facial feature spaces through constrained feature learning and introduces a gradient suppression algorithm to filter abnormal gradients, preventing shortcut learning and enhancing generalisable feature extraction. Experiments on hybrid datasets validate DIRE's effectiveness in four cross-domain tasks (O&C&M→I, O&C&I→M, O&I&M→C, and I&C&M→O). Ablation studies confirm the synergy of feature standardisation and gradient suppression, reducing bias by 97.6% and parameters by 42%, while accelerating inference by 2.3×. DIRE achieves 98.2% and 96.7% accuracy on two tasks (O&C&I→M and O&M&I→C), outperforming state-of-the-art methods by 5.3% while maintaining computational efficiency. This study advances generative face detection through dual optimisation, offering a lightweight framework for financial identity verification and social media content moderation.

Keywords: denoising diffusion probabilistic models; DDPMs; diffusion reconstruction error; DIRE; face anti-counterfeiting; disentanglement.

DOI: 10.1504/IJICT.2025.146364

International Journal of Information and Communication Technology, 2025 Vol.26 No.14, pp.20 - 43

Received: 23 Dec 2024
Accepted: 27 Mar 2025
Published online: 27 May 2025 *

Title: Diffusion-generated face image identification technique based on diffusion reconstruction error feature

Keep up-to-date