Rethinking Cross-Generator Image Forgery Detection through DINOv3

TL;DR

This paper demonstrates that frozen visual foundation models like DINOv3 can effectively detect cross-generator image forgeries by leveraging global low-frequency cues, without additional training, providing a universal and interpretable detection baseline.

Contribution

It reveals that DINOv3 inherently captures transferable authenticity cues and introduces a simple token-ranking method to enhance detection accuracy without fine-tuning.

Findings

DINOv3 shows strong cross-generator detection ability without fine-tuning.

Global, low-frequency structures serve as transferable authenticity cues.

Token selection improves detection accuracy across datasets.

Abstract

As generative models become increasingly diverse and powerful, cross-generator detection has emerged as a new challenge. Existing detection methods often memorize artifacts of specific generative models rather than learning transferable cues, leading to substantial failures on unseen generators. Surprisingly, this work finds that frozen visual foundation models, especially DINOv3, already exhibit strong cross-generator detection capability without any fine-tuning. Through systematic studies on frequency, spatial, and token perspectives, we observe that DINOv3 tends to rely on global, low-frequency structures as weak but transferable authenticity cues instead of high-frequency, generator-specific artifacts. Motivated by this insight, we introduce a simple, training-free token-ranking strategy followed by a lightweight linear probe to select a small subset of authenticity-relevant tokens. This token subset consistently improves detection accuracy across all evaluated datasets. Our study provides empirical evidence and a feasible hypothesis for understanding why foundation models generalize across diverse generators, offering a universal, efficient, and interpretable baseline for image forgery detection.