Smoothing the Disentangled Latent Style Space for Unsupervised Image-to-Image Translation

TL;DR

This paper introduces a new training protocol with specific losses to create a smooth, disentangled latent style space in image-to-image translation models, improving interpolation quality and content preservation.

Contribution

It proposes a novel training method that enhances the smoothness and disentanglement of the latent style space in I2I translation models, applicable to existing approaches.

Findings

Significantly improves image quality during translation.

Enhances the smoothness of interpolations across and within domains.

Better preserves source image content during translation.

Abstract

Image-to-Image (I2I) multi-domain translation models are usually evaluated also using the quality of their semantic interpolation results. However, state-of-the-art models frequently show abrupt changes in the image appearance during interpolation, and usually perform poorly in interpolations across domains. In this paper, we propose a new training protocol based on three specific losses which help a translation network to learn a smooth and disentangled latent style space in which: 1) Both intra- and inter-domain interpolations correspond to gradual changes in the generated images and 2) The content of the source image is better preserved during the translation. Moreover, we propose a novel evaluation metric to properly measure the smoothness of latent style space of I2I translation models. The proposed method can be plugged into existing translation approaches, and our extensive experiments on different datasets show that it can significantly boost the quality of the generated images and the graduality of the interpolations.