Reduce the Artifacts Bias for More Generalizable AI-Generated Image Detection

2026-05-14

AI Summary

This work addresses the limited generalizability of existing AI-generated image detection methods, which often rely on a single reconstruction technique and consequently introduce homogeneous artifacts that fail to capture the diverse characteristics of generative models such as GANs. To overcome this, the authors propose a Separate Expert Fusion framework that integrates complementary forgery artifacts derived from both VAE/DDIM-based reconstruction and GAN-based upsampling, while preserving alignment in content, size, and format. Domain-specific experts are constructed by freezing foundational models and applying LoRA-based fine-tuning, and their features are fused through a gating mechanism to enable decoupled yet synergistic representation learning. This approach effectively mitigates inter-domain interference and achieves significantly improved detection generalization across 13 diverse benchmarks for both diffusion- and GAN-generated images.

Abstract

As the misuse of AI-generated images grows, generalizable image detection techniques are urgently needed. Recent state-of-the-art (SOTA) methods adopt aligned training datasets to reduce content, size, and format biases, empowering models to capture robust forgery cues. A common strategy is to employ reconstruction techniques, e.g., VAE and DDIM, which show remarkable results in diffusion-based methods. However, such reconstruction-based approaches typically introduce limited and homogeneous artifacts, which cannot fully capture diverse generative patterns, such as GAN-based methods. To complement reconstruction-based fake images with aligned yet diverse artifact patterns, we propose a GAN-based upsampling approach that mimics GAN-generated fake patterns while preserving content, size, and format alignment. This naturally results in two aligned but distinct types of fake images. However, due to the domain shift between reconstruction-based and upsampling-based fake images, direct mixed training causes suboptimal results, where one domain disrupts feature learning of the other. Accordingly, we propose a Separate Expert Fusion (SEF) framework to extract complementary artifact information and reduce inter-domain interference. We first train domain-specific experts via LoRA adaptation on a frozen foundational model, then conduct decoupled fusion with a gating network to adaptively combine expert features while retaining their specialized knowledge. Rather than merely benefiting GAN-generated image detection, this design introduces diverse and complementary artifact patterns that enable SEF to learn a more robust decision boundary and improve generalization across broader generative methods. Extensive experiments demonstrate that our method yields strong results across 13 diverse benchmarks. Codes are released at: https://github.com/liyih/SEF_AIGC_detection.

Problem

• AI-generated image detection
• artifacts bias
• generalization
• GAN
• diffusion models

Innovation

• Separate Expert Fusion
• GAN-based upsampling
• artifact diversity
• domain-specific experts
• generalizable detection