Breaking Semantic Artifacts for Generalized AI-generated Image Detection

Chende Zheng, Chenhao Lin, Zhengyu Zhao, Hang Wang, Xu Guo, Shuai Liu, Chao Shen

Keywords: AI Security; Deepfake Detection; AI-generated Image; Diffusion Model;

TL;DR: The paper introduces "semantic artifacts" in cross-scene images, which leads to performance drops on generalized detection. A dataset from various generative models and scenes is built and a simple yet effective patch-based method is proposed.

Abstract:

With the continuous evolution of AI-generated images, the generalized detection of them has become a crucial aspect of AI security.
Existing detectors have focused on cross-generator generalization, while it remains unexplored whether these detectors can generalize across different image scenes, e.g., images from different datasets with different semantics. In this paper, we reveal that existing detectors suffer from substantial Accuracy drops in such cross-scene generalization. In particular, we attribute their failures to ''semantic artifacts'' in both real and generated images, to which detectors may overfit. To break such ''semantic artifacts'', we propose a simple yet effective approach based on conducting an image patch shuffle and then training an end-to-end patch-based classifier. We conduct a comprehensive open-world evaluation on 31 test sets, covering 7 Generative Adversarial Networks, 18 (variants of) Diffusion Models, and another 6 CNN-based generative models. The results demonstrate that our approach outperforms previous approaches by 2.08\% (absolute) on average regarding cross-scene detection Accuracy. We also notice the superiority of our approach in open-world generalization, with an average Accuracy improvement of 10.59\% (absolute) across all test sets. Our code is available at *https://github.com/Zig-HS/FakeImageDetection*.

Primary Area: Other (please use sparingly, only use the keyword field for more details)

Submission Number: 6879