Few-Shot Learner Generalizes Across AI-Generated Image Detection

TL;DR

This paper introduces Few-Shot Detector (FSD), a novel method that effectively detects unseen AI-generated fake images using minimal samples by learning a specialized metric space, outperforming existing detectors.

Contribution

The paper presents FSD, a new few-shot learning approach for AI-generated image detection that generalizes well to unseen models with minimal training data.

Findings

FSD achieves +11.6% accuracy on GenImage dataset with only 10 samples

FSD better captures intra-category commonality in unseen images

FSD outperforms existing detectors in zero-shot scenarios

Abstract

Current fake image detectors trained on large synthetic image datasets perform satisfactorily on limited studied generative models. However, these detectors suffer a notable performance decline over unseen models. Besides, collecting adequate training data from online generative models is often expensive or infeasible. To overcome these issues, we propose Few-Shot Detector (FSD), a novel AI-generated image detector which learns a specialized metric space for effectively distinguishing unseen fake images using very few samples. Experiments show that FSD achieves state-of-the-art performance by $+11.6\%$ average accuracy on the GenImage dataset with only $10$ additional samples. More importantly, our method is better capable of capturing the intra-category commonality in unseen images without further training. Our code is available at https://github.com/teheperinko541/Few-Shot-AIGI-Detector.