Generator Knows What Discriminator Should Learn in Unconditional GANs

TL;DR

This paper introduces a novel generator-guided discriminator regularization (GGDR) method for unconditional GANs, using generator feature maps to enhance discriminator learning and improve image generation quality.

Contribution

It proposes a new regularization technique that leverages generator feature maps to guide discriminator training in unconditional GANs, reducing reliance on semantic labels.

Findings

GGDR improves unconditional GAN performance across multiple datasets.

The method enhances both quantitative metrics and qualitative image quality.

Discriminator learns richer semantic representations with GGDR.

Abstract

Recent methods for conditional image generation benefit from dense supervision such as segmentation label maps to achieve high-fidelity. However, it is rarely explored to employ dense supervision for unconditional image generation. Here we explore the efficacy of dense supervision in unconditional generation and find generator feature maps can be an alternative of cost-expensive semantic label maps. From our empirical evidences, we propose a new generator-guided discriminator regularization(GGDR) in which the generator feature maps supervise the discriminator to have rich semantic representations in unconditional generation. In specific, we employ an U-Net architecture for discriminator, which is trained to predict the generator feature maps given fake images as inputs. Extensive experiments on mulitple datasets show that our GGDR consistently improves the performance of baseline methods in terms of quantitative and qualitative aspects. Code is available at https://github.com/naver-ai/GGDR