Detecting Images Generated by Deep Diffusion Models using their Local Intrinsic Dimensionality

TL;DR

This paper introduces a lightweight method based on Local Intrinsic Dimensionality to effectively detect images generated by diffusion models, outperforming existing techniques and establishing a comprehensive benchmark.

Contribution

The paper presents a novel multiLID approach for detecting diffusion-generated images and identifying their source models, demonstrating superior performance over prior methods.

Findings

multiLID achieves near-perfect detection accuracy

The method outperforms existing detection approaches

A new comprehensive benchmark for diffusion image detection is established

Abstract

Diffusion models recently have been successfully applied for the visual synthesis of strikingly realistic appearing images. This raises strong concerns about their potential for malicious purposes. In this paper, we propose using the lightweight multi Local Intrinsic Dimensionality (multiLID), which has been originally developed in context of the detection of adversarial examples, for the automatic detection of synthetic images and the identification of the according generator networks. In contrast to many existing detection approaches, which often only work for GAN-generated images, the proposed method provides close to perfect detection results in many realistic use cases. Extensive experiments on known and newly created datasets demonstrate that the proposed multiLID approach exhibits superiority in diffusion detection and model identification. Since the empirical evaluations of recent publications on the detection of generated images are often mainly focused on the "LSUN-Bedroom" dataset, we further establish a comprehensive benchmark for the detection of diffusion-generated images, including samples from several diffusion models with different image sizes.