English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Unpaired image-to-image translation has broad applications in art, design,and scientific simulations. One early breakthrough was CycleGAN that emphasizesone-to-one mappings between two unpaired image domains viagenerative-adversarial networks (GAN) coupled with the cycle-consistencyconstraint, while more recent works promote one-to-many mapping to boostdiversity of the translated images. Motivated by scientific simulation andone-to-one needs, this work revisits the classic CycleGAN framework and boostsits performance to outperform more contemporary models without relaxing thecycle-consistency constraint. To achieve this, we equip the generator with aVision Transformer (ViT) and employ necessary training and regularizationtechniques. Compared to previous best-performing models, our model performsbetter and retains a strong correlation between the original and translatedimage. An accompanying ablation study shows that both the gradient penalty andself-supervised pre-training are crucial to the improvement. To promotereproducibility and open science, the source code, hyperparameterconfigurations, and pre-trained model are available athttps://github.com/LS4GAN/uvcgan.

UVCGAN: UNet Vision Transformer cycle-consistent GAN for unpaired image-to-image translation