LaRE2: Latent Reconstruction Error Based Method for Diffusion-Generated Image Detection

2024-03-26

Computer Vision and Pattern Recognition

AI Summary

The rapid advancement of diffusion-based image generation has rendered synthetic images increasingly indistinguishable from real ones, posing severe privacy and security threats. To address this, we propose LaRE², a latent-space reconstruction error–based detection method. LaRE² is the first to define reconstruction error in the latent space as a discriminative feature and introduces an Error-Guided Reconstruction Enhancement (EGRE) module that jointly optimizes spatial and channel dimensions via a “first-align-then-refine” strategy. Crucially, LaRE² operates without fine-tuning the generative models, ensuring strong generalizability across diverse generators. Evaluated on the GenImage benchmark, LaRE² achieves state-of-the-art performance, outperforming prior methods by 11.9% in average accuracy (ACC) and 12.1% in average precision (AP), while accelerating feature extraction by 8×.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Security and Privacy: Large-scale security measurements
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

The evolution of Diffusion Models has dramatically improved image generation quality, making it increasingly difficult to differentiate between real and generated images. This development, while impressive, also raises significant privacy and security concerns. In response to this, we propose a novel Latent REconstruction error guided feature REfinement method (LaRE2) for detecting the diffusion-generated images. We come up with the Latent Reconstruction Error (LaRE), the first reconstruction-error based feature in the latent space for generated image detection. LaRE surpasses existing methods in terms of feature extraction efficiency while preserving crucial cues required to differentiate between the real and the fake. To exploit LaRE, we propose an Error-Guided feature REfinement module (EGRE), which can refine the image feature guided by LaRE to enhance the discriminativeness of the feature. Our EGRE utilizes an align-then-refine mechanism, which effectively refines the image feature for generated-image detection from both spatial and channel perspectives. Extensive experiments on the large-scale GenImage benchmark demonstrate the superiority of our LaRE2, which surpasses the best SoTA method by up to 11.9%/12.1% average ACC/AP across 8 different image generators. LaRE also surpasses existing methods in terms of feature extraction cost, delivering an impressive speed enhancement of 8 times.

Problem

• Detects diffusion-generated images
• Enhances feature extraction efficiency
• Improves image feature discriminativeness

Innovation

• Latent Reconstruction Error feature
• Error-Guided feature Refinement module
• Align-then-refine mechanism