Detecting GAN-generated Imagery using Color Cues

TL;DR

This paper investigates the structural differences in color processing between GAN-generated images and real camera images, proposing a method to detect synthetic images based on these color cues.

Contribution

It introduces a novel approach to detect GAN-generated images by analyzing their color treatment, highlighting differences from real camera images.

Findings

Color cues effectively distinguish GAN images from real images

GANs treat color differently than real cameras in specific ways

Proposed method achieves accurate detection of GAN imagery

Abstract

Image forensics is an increasingly relevant problem, as it can potentially address online disinformation campaigns and mitigate problematic aspects of social media. Of particular interest, given its recent successes, is the detection of imagery produced by Generative Adversarial Networks (GANs), e.g. `deepfakes'. Leveraging large training sets and extensive computing resources, recent work has shown that GANs can be trained to generate synthetic imagery which is (in some ways) indistinguishable from real imagery. We analyze the structure of the generating network of a popular GAN implementation, and show that the network's treatment of color is markedly different from a real camera in two ways. We further show that these two cues can be used to distinguish GAN-generated imagery from camera imagery, demonstrating effective discrimination between GAN imagery and real camera images used to train the GAN.