Community Forensics: Using Thousands of Generators to Train Fake Image Detectors

TL;DR

This paper introduces a large, diverse dataset of 2.7 million images from thousands of generative models to improve the detection of AI-generated images, demonstrating that increased diversity enhances detector performance.

Contribution

The creation of a significantly larger and more diverse dataset of AI-generated images from thousands of models, enabling better generalization of fake image detectors.

Findings

Detection improves with more models in training.

Diversity of models enhances detection accuracy.

Detectors trained on this dataset outperform those trained on previous datasets.

Abstract

One of the key challenges of detecting AI-generated images is spotting images that have been created by previously unseen generative models. We argue that the limited diversity of the training data is a major obstacle to addressing this problem, and we propose a new dataset that is significantly larger and more diverse than prior work. As part of creating this dataset, we systematically download thousands of text-to-image latent diffusion models and sample images from them. We also collect images from dozens of popular open source and commercial models. The resulting dataset contains 2.7M images that have been sampled from 4803 different models. These images collectively capture a wide range of scene content, generator architectures, and image processing settings. Using this dataset, we study the generalization abilities of fake image detectors. Our experiments suggest that detection performance improves as the number of models in the training set increases, even when these models have similar architectures. We also find that detection performance improves as the diversity of the models increases, and that our trained detectors generalize better than those trained on other datasets. The dataset can be found in https://jespark.net/projects/2024/community_forensics