Decoupling Semantics and Fingerprints: A Universal Representation for AI-Generated Image Detection

TL;DR

This paper introduces ODP-Net, a novel method that disentangles universal forgery traces from generator-specific fingerprints and semantics, improving AI-generated image detection across unseen architectures.

Contribution

It proposes a spectral analysis-based orthogonal decomposition approach to structurally disentangle factors, enabling better generalization in forgery detection.

Findings

ODP-Net outperforms existing methods on unseen architectures.

Spectral analysis reveals disjoint frequency subspaces for different features.

Explicit disentanglement improves detection robustness.

Abstract

Detecting AI-generated images across unseen architectures remains challenging, as existing models often overfit to generator-specific fingerprints and semantic content rather than learning universal forgery traces. We attribute this failure to feature entanglement: detectors learn these factors as a single entangled representation, where universal forgery traces are inextricably confounded with both generator-specific fingerprints and semantic content. Crucially, our spectral analysis reveals that this entanglement is avoidable: distinct generator-specific fingerprints (e.g., GAN stripes vs. Diffusion Model spots) occupy disjoint frequency subspaces and coexist as independent superpositions. Leveraging this physical orthogonality, we propose the Orthogonal Decomposition and Purification Network (ODP-Net) to structurally disentangle these factors. Specifically, ODP-Net employs (1) Instance-aware Orthogonal Decomposition to project features into mutually exclusive subspaces: universal forgery traces, generator-specific fingerprints, and semantic content; (2) Perturbation-based Purification to enforce semantic invariance via cross-sample feature injection; and (3) Manifold Alignment to bridge domain gaps. By explicitly decoupling universal forgery traces from generator-specific fingerprints and semantic content, ODP-Net achieves state-of-the-art performance on unseen architectures (e.g., Stable Diffusion 3), validating that structural disentanglement is key to generalization.