RepMix: Representation Mixing for Robust Attribution of Synthesized Images
This addresses the challenge of detecting and tracing GAN-generated images for security and forensic applications, representing a strong specific gain in the domain.
The paper tackles the problem of attributing synthetic images to their GAN architectures, invariant to semantic content and robust to common transformations, by proposing RepMix and introducing the Attribution88 benchmark, achieving significant improvements over existing methods.
Rapid advances in Generative Adversarial Networks (GANs) raise new challenges for image attribution; detecting whether an image is synthetic and, if so, determining which GAN architecture created it. Uniquely, we present a solution to this task capable of 1) matching images invariant to their semantic content; 2) robust to benign transformations (changes in quality, resolution, shape, etc.) commonly encountered as images are re-shared online. In order to formalize our research, a challenging benchmark, Attribution88, is collected for robust and practical image attribution. We then propose RepMix, our GAN fingerprinting technique based on representation mixing and a novel loss. We validate its capability of tracing the provenance of GAN-generated images invariant to the semantic content of the image and also robust to perturbations. We show our approach improves significantly from existing GAN fingerprinting works on both semantic generalization and robustness. Data and code are available at https://github.com/TuBui/image_attribution.