SFLD: Reducing the content bias for AI-generated Image Detection

TL;DR

This paper introduces SFLD, a robust method combining PatchShuffle with high-level and low-level features to improve AI-generated image detection, and presents TwinSynths, a new benchmark for evaluating detection methods.

Contribution

We propose SFLD, a novel detection approach that enhances robustness and generalization, and introduce TwinSynths, a high-quality benchmark for evaluating AI-generated image detectors.

Findings

SFLD outperforms existing detectors across various generative models.

PatchShuffle improves robustness against image degradations.

TwinSynths provides high-quality, diverse synthetic images for benchmarking.

Abstract

Identifying AI-generated content is critical for the safe and ethical use of generative AI. Recent research has focused on developing detectors that generalize to unknown generators, with popular methods relying either on high-level features or low-level fingerprints. However, these methods have clear limitations: biased towards unseen content, or vulnerable to common image degradations, such as JPEG compression. To address these issues, we propose a novel approach, SFLD, which incorporates PatchShuffle to integrate high-level semantic and low-level textural information. SFLD applies PatchShuffle at multiple levels, improving robustness and generalization across various generative models. Additionally, current benchmarks face challenges such as low image quality, insufficient content preservation, and limited class diversity. In response, we introduce TwinSynths, a new benchmark generation methodology that constructs visually near-identical pairs of real and synthetic images to ensure high quality and content preservation. Our extensive experiments and analysis show that SFLD outperforms existing methods on detecting a wide variety of fake images sourced from GANs, diffusion models, and TwinSynths, demonstrating the state-of-the-art performance and generalization capabilities to novel generative models.