CNN-generated images are surprisingly easy to spot... for now

Abstract

In this work we ask whether it is possible to create a "universal" detector for telling apart real images from these generated by a CNN, regardless of architecture or dataset used. To test this, we collect a dataset consisting of fake images generated by 11 different CNN-based image generator models, chosen to span the space of commonly used architectures today (ProGAN, StyleGAN, BigGAN, CycleGAN, StarGAN, GauGAN, DeepFakes, cascaded refinement networks, implicit maximum likelihood estimation, second-order attention super-resolution, seeing-in-the-dark). We demonstrate that, with careful pre- and post-processing and data augmentation, a standard image classifier trained on only one specific CNN generator (ProGAN) is able to generalize surprisingly well to unseen architectures, datasets, and training methods (including the just released StyleGAN2). Our findings suggest the intriguing possibility that today's CNN-generated images share some common systematic flaws, preventing them from achieving realistic image synthesis. Code and pre-trained networks are available at https://peterwang512.github.io/CNNDetection/ .

arXiv commentAccepted to CVPR 2020

References in corpus (5)

• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Self-Attention Generative Adversarial Networks
• LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop
• ForensicTransfer: Weakly-supervised Domain Adaptation for Forgery Detection
• Why do deep convolutional networks generalize so poorly to small image transformations?

Cited by in corpus (3)

• Comprehensive Dataset of Synthetic and Manipulated Overhead Imagery for Development and Evaluation of Forensic Tools
• On the Theoretical Equivalence of Several Trade-Off Curves Assessing Statistical Proximity
• On the Frequency Bias of Generative Models