Image Augmentations for GAN Training

TL;DR

This paper systematically investigates the use of image augmentations in GAN training, revealing that augmentations significantly enhance image quality and can achieve state-of-the-art results when combined with regularizations.

Contribution

It provides comprehensive insights and guidelines on applying image augmentations to improve GAN training and introduces new state-of-the-art results for conditional image generation.

Findings

Augmentations enable vanilla GANs to match state-of-the-art quality.

Combining augmentations with regularizations further improves results.

Achieved new state-of-the-art on CIFAR-10 with regularizations.

Abstract

Data augmentations have been widely studied to improve the accuracy and robustness of classifiers. However, the potential of image augmentation in improving GAN models for image synthesis has not been thoroughly investigated in previous studies. In this work, we systematically study the effectiveness of various existing augmentation techniques for GAN training in a variety of settings. We provide insights and guidelines on how to augment images for both vanilla GANs and GANs with regularizations, improving the fidelity of the generated images substantially. Surprisingly, we find that vanilla GANs attain generation quality on par with recent state-of-the-art results if we use augmentations on both real and generated images. When this GAN training is combined with other augmentation-based regularization techniques, such as contrastive loss and consistency regularization, the augmentations further improve the quality of generated images. We provide new state-of-the-art results for conditional generation on CIFAR-10 with both consistency loss and contrastive loss as additional regularizations.