Biphasic Learning of GANs for High-Resolution Image-to-Image Translation

TL;DR

This paper introduces biphasic learning, a novel training framework for GANs that stabilizes high-resolution image-to-image translation, improving quality and training stability across multiple visual domains.

Contribution

The paper proposes a biphasic learning framework with an adjustable objective, inherited adversarial loss, and perceptual consistency loss for high-resolution GAN training.

Findings

Significantly outperforms existing methods in quantitative metrics.

Achieves stable training and high-quality results at 1024^2 resolution.

Effective across various face-related synthesis tasks.

Abstract

Despite that the performance of image-to-image translation has been significantly improved by recent progress in generative models, current methods still suffer from severe degradation in training stability and sample quality when applied to the high-resolution situation. In this work, we present a novel training framework for GANs, namely biphasic learning, to achieve image-to-image translation in multiple visual domains at $1024^2$ resolution. Our core idea is to design an adjustable objective function that varies across training phases. Within the biphasic learning framework, we propose a novel inherited adversarial loss to achieve the enhancement of model capacity and stabilize the training phase transition. Furthermore, we introduce a perceptual-level consistency loss through mutual information estimation and maximization. To verify the superiority of the proposed method, we apply it to a wide range of face-related synthesis tasks and conduct experiments on multiple large-scale datasets. Through comprehensive quantitative analyses, we demonstrate that our method significantly outperforms existing methods.