MIRROR: Manifold Ideal Reference ReconstructOR for Generalizable AI-Generated Image Detection

2026-02-02

AI Summary

This work addresses the limited generalizability of current AI-generated image detection methods, which rely on classifying artifacts specific to particular generative models and thus struggle to keep pace with their rapid evolution. To overcome this, the authors reformulate detection as a reference consistency verification problem: an ideal reference image conforming to the real-image manifold is constructed, and the reconstruction residual between the input and this reference serves as the discriminative signal. The core innovation lies in introducing a detection paradigm grounded in the human cognitive manifold, integrating a learnable discrete memory bank to encode real-world priors and generating manifold-consistent references via sparse linear combinations atop a pretrained backbone network. The method achieves state-of-the-art performance across 14 benchmarks, improving accuracy by 2.1% on standard datasets and 8.1% on in-the-wild sets, and attains 89.6% accuracy on the Human-AIGI benchmark—significantly surpassing both laypersons and vision experts.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Humans and AI: Other Foundations of Human Computation & AI
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

High-fidelity generative models have narrowed the perceptual gap between synthetic and real images, posing serious threats to media security. Most existing AI-generated image (AIGI) detectors rely on artifact-based classification and struggle to generalize to evolving generative traces. In contrast, human judgment relies on stable real-world regularities, with deviations from the human cognitive manifold serving as a more generalizable signal of forgery. Motivated by this insight, we reformulate AIGI detection as a Reference-Comparison problem that verifies consistency with the real-image manifold rather than fitting specific forgery cues. We propose MIRROR (Manifold Ideal Reference ReconstructOR), a framework that explicitly encodes reality priors using a learnable discrete memory bank. MIRROR projects an input into a manifold-consistent ideal reference via sparse linear combination, and uses the resulting residuals as robust detection signals. To evaluate whether detectors reach the"superhuman crossover"required to replace human experts, we introduce the Human-AIGI benchmark, featuring a psychophysically curated human-imperceptible subset. Across 14 benchmarks, MIRROR consistently outperforms prior methods, achieving gains of 2.1% on six standard benchmarks and 8.1% on seven in-the-wild benchmarks. On Human-AIGI, MIRROR reaches 89.6% accuracy across 27 generators, surpassing both lay users and visual experts, and further approaching the human perceptual limit as pretrained backbones scale. The code is publicly available at: https://github.com/349793927/MIRROR

Problem

• AI-generated image detection
• generalization
• media security
• real-image manifold
• forgery detection

Innovation

• manifold learning
• reference-comparison framework
• discrete memory bank
• generalizable detection
• AI-generated image forensics