Deepfake Network Architecture Attribution

TL;DR

This paper introduces DNA-Det, a novel method for attributing fake images to their source GAN architectures, capable of handling models with different training configurations and outperforming existing model-level attribution techniques.

Contribution

The paper presents the first approach for architecture-level deepfake attribution, addressing limitations of model-level methods in real-world scenarios with finetuned or retrained models.

Findings

DNA-Det effectively attributes fake images to their source architectures.

The method performs well across multiple cross-test setups.

DNA-Det outperforms existing model-level attribution techniques.

Abstract

With the rapid progress of generation technology, it has become necessary to attribute the origin of fake images. Existing works on fake image attribution perform multi-class classification on several Generative Adversarial Network (GAN) models and obtain high accuracies. While encouraging, these works are restricted to model-level attribution, only capable of handling images generated by seen models with a specific seed, loss and dataset, which is limited in real-world scenarios when fake images may be generated by privately trained models. This motivates us to ask whether it is possible to attribute fake images to the source models' architectures even if they are finetuned or retrained under different configurations. In this work, we present the first study on Deepfake Network Architecture Attribution to attribute fake images on architecture-level. Based on an observation that GAN architecture is likely to leave globally consistent fingerprints while traces left by model weights vary in different regions, we provide a simple yet effective solution named DNA-Det for this problem. Extensive experiments on multiple cross-test setups and a large-scale dataset demonstrate the effectiveness of DNA-Det.