FakeInversion: Learning to Detect Images from Unseen Text-to-Image Models by Inverting Stable Diffusion

TL;DR

This paper introduces FakeInversion, a novel synthetic image detector leveraging inversion features from Stable Diffusion, which generalizes effectively to unseen high-fidelity generators and sets new state-of-the-art performance.

Contribution

The paper presents a new detection method based on inversion features from Stable Diffusion, improving generalization to unseen models and establishing a robust evaluation protocol with public benchmarks.

Findings

Achieves state-of-the-art detection accuracy across multiple setups

Generalizes well to unseen high-fidelity generators like DALL-E 3

Introduces a new evaluation protocol using reverse image search

Abstract

Due to the high potential for abuse of GenAI systems, the task of detecting synthetic images has recently become of great interest to the research community. Unfortunately, existing image-space detectors quickly become obsolete as new high-fidelity text-to-image models are developed at blinding speed. In this work, we propose a new synthetic image detector that uses features obtained by inverting an open-source pre-trained Stable Diffusion model. We show that these inversion features enable our detector to generalize well to unseen generators of high visual fidelity (e.g., DALL-E 3) even when the detector is trained only on lower fidelity fake images generated via Stable Diffusion. This detector achieves new state-of-the-art across multiple training and evaluation setups. Moreover, we introduce a new challenging evaluation protocol that uses reverse image search to mitigate stylistic and thematic biases in the detector evaluation. We show that the resulting evaluation scores align well with detectors' in-the-wild performance, and release these datasets as public benchmarks for future research.