Unsupervised multi-domain image translation with domain representation learning

作者：

Highlights：

• A novel DoReGAN is proposed to achieve multi-domain image translation;

• A representation loss is presented to automatically extract domain-specific representation vector.

• Extensive experiments validate the effect of our DoReGAN quantitatively and qualitatively.

摘要

•A novel DoReGAN is proposed to achieve multi-domain image translation;•A representation loss is presented to automatically extract domain-specific representation vector.•Extensive experiments validate the effect of our DoReGAN quantitatively and qualitatively.

论文关键词：Image-to-image translation,Multi-domain,Unsupervised,Attribute label

论文评审过程：Received 25 October 2019, Revised 14 December 2020, Accepted 23 August 2021, Available online 30 August 2021, Version of Record 6 September 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116452