Model Discrepancy Learning: Synthetic Faces Detection Based on Multi-Reconstruction

2025-04-10

AI Summary

Existing synthetic face detection methods inadequately address the growing misuse of generative AI, as they overlook fundamental architectural and statistical disparities among generation techniques—particularly between GANs and diffusion models—resulting in poor generalization across models and domains. Method: We propose a multi-reconstruction discrepancy-based detection framework that (i) explicitly models the intrinsic alignment between generative architectures and their outputs, establishing a novel model-difference learning paradigm; (ii) extracts robust reconstruction residual features via joint inverse reconstruction across multiple pre-trained generators; and (iii) enhances robustness through cross-architecture consistency analysis. Contribution/Results: We introduce ASFD—the first large-scale, Asian-centric synthetic face dataset covering diverse generation techniques. Extensive experiments demonstrate state-of-the-art performance under cross-model, cross-dataset, and noise-corrupted settings, achieving superior generalization and fine-grained generation-technology attribution capability.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Advances in image generation enable hyper-realistic synthetic faces but also pose risks, thus making synthetic face detection crucial. Previous research focuses on the general differences between generated images and real images, often overlooking the discrepancies among various generative techniques. In this paper, we explore the intrinsic relationship between synthetic images and their corresponding generation technologies. We find that specific images exhibit significant reconstruction discrepancies across different generative methods and that matching generation techniques provide more accurate reconstructions. Based on this insight, we propose a Multi-Reconstruction-based detector. By reversing and reconstructing images using multiple generative models, we analyze the reconstruction differences among real, GAN-generated, and DM-generated images to facilitate effective differentiation. Additionally, we introduce the Asian Synthetic Face Dataset (ASFD), containing synthetic Asian faces generated with various GANs and DMs. This dataset complements existing synthetic face datasets. Experimental results demonstrate that our detector achieves exceptional performance, with strong generalization and robustness.

Problem

• Detect synthetic faces by analyzing reconstruction discrepancies
• Explore intrinsic links between synthetic images and generation techniques
• Address lack of diverse generative method representation in datasets

Innovation

• Multi-Reconstruction-based detector for synthetic faces
• Analyzes reconstruction differences across generative models
• Introduces Asian Synthetic Face Dataset (ASFD)