Abstract:AI-generated image detectors suffer significant performance degradation under real-world image corruptions such as JPEG compression, Gaussian blur, and resolution downsampling. We observe that state-of-the-art methods, including B-Free, treat degradation robustness as a byproduct of data augmentation rather than an explicit training objective. In this work, we propose Degradation-Consistent Paired Training (DCPT), a simple yet effective training strategy that explicitly enforces robustness through paired consistency constraints. For each training image, we construct a clean view and a degraded view, then impose two constraints: a feature consistency loss that minimizes the cosine distance between clean and degraded representations, and a prediction consistency loss based on symmetric KL divergence that aligns output distributions across views. DCPT adds zero additional parameters and zero inference overhead. Experiments on the Synthbuster benchmark (9 generators, 8 degradation conditions) demonstrate that DCPT improves the degraded-condition average accuracy by 9.1 percentage points compared to an identical baseline without paired training, while sacrificing only 0.9% clean accuracy. The improvement is most pronounced under JPEG compression (+15.7% to +17.9%). Ablation further reveals that adding architectural components leads to overfitting on limited training data, confirming that training objective improvement is more effective than architectural augmentation for degradation robustness.
| Comments: | 6 pages, 5 figures, 2 tables |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2604.10102 [cs.CV] |
| (or arXiv:2604.10102v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2604.10102 arXiv-issued DOI via DataCite |
From: Zongyou Yang [view email]
[v1]
Sat, 11 Apr 2026 08:52:28 UTC (499 KB)
[v2]
Tue, 26 May 2026 15:45:17 UTC (499 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。