An Optimized Architecture for Unpaired Image-to-Image Translation

TL;DR

This paper presents an optimized unpaired image-to-image translation architecture that simplifies training by focusing only on translating from domain A to B and introduces a new loss to improve efficiency and reduce training time.

Contribution

It proposes a novel architecture that eliminates the reverse mapping in Cycle-GAN and introduces a Deviation-loss, leading to faster training and improved performance.

Findings

Significantly reduced training duration

Effective translation from domain A to B without reverse mapping

Improved image translation quality

Abstract

Unpaired Image-to-Image translation aims to convert the image from one domain (input domain A) to another domain (target domain B), without providing paired examples for the training. The state-of-the-art, Cycle-GAN demonstrated the power of Generative Adversarial Networks with Cycle-Consistency Loss. While its results are promising, there is scope for optimization in the training process. This paper introduces a new neural network architecture, which only learns the translation from domain A to B and eliminates the need for reverse mapping (B to A), by introducing a new Deviation-loss term. Furthermore, few other improvements to the Cycle-GAN are found and utilized in this new architecture, contributing to significantly lesser training duration.