Recent Advances on Generalizable Diffusion-generated Image Detection

TL;DR

This work tackles the rising challenge of detecting diffusion-generated images with strong cross-model generalization. It presents a taxonomy that separates methods into data-driven and feature-driven families, further dividing each into fine-grained categories, and it surveys concrete techniques from advanced architectures to reconstruction-based and likelihood-driven approaches. The paper identifies core open problems—robustness to post-processing, theoretical grounding, dataset quality, and alternative detection paradigms—and offers directions such as mixture-of-experts to improve real-world performance. Overall, it provides a structured map of current methods, practical insights for deployment, and a roadmap for future research in diffusion-generated image detection.

Abstract

The rise of diffusion models has significantly improved the fidelity and diversity of generated images. With numerous benefits, these advancements also introduce new risks. Diffusion models can be exploited to create high-quality Deepfake images, which poses challenges for image authenticity verification. In recent years, research on generalizable diffusion-generated image detection has grown rapidly. However, a comprehensive review of this topic is still lacking. To bridge this gap, we present a systematic survey of recent advances and classify them into two main categories: (1) data-driven detection and (2) feature-driven detection. Existing detection methods are further classified into six fine-grained categories based on their underlying principles. Finally, we identify several open challenges and envision some future directions, with the hope of inspiring more research work on this important topic. Reviewed works in this survey can be found at https://github.com/zju-pi/Awesome-Diffusion-generated-Image-Detection.

Figures (2)

• Figure 1: Generated images can now easily mislead the public, leading to serious consequences such as panic and economic losses.
• Figure 2: A taxonomy of recent diffusion-generated image detection methods.