Artificial Intelligence


Improved Multi-Domain Image-to-Image Translation GAN

Authors: Jeongik Cho

StarGAN has shown excellent performance in image-to-image translation using adversarial, reconstruction, and classification losses in multi-domain image-to-image translation. The Style-Based Generator Architecture boosts generator performance through the Embedder and AdaIn modules. I propose here an attribute loss, which is like having multiple GANs, which is enhanced by combining StarGAN's conditional GAN loss (adversarial loss and classification loss) to improve learning speed. And suggest the new generator architecture, whose name is bi-directional progressive growing Style-Based U-Net generator, to improve learning speed.

Comments: 7 Pages.

Download: PDF

Submission history

[v1] 2019-09-03 20:56:00

Unique-IP document downloads: 7 times is a pre-print repository rather than a journal. Articles hosted may not yet have been verified by peer-review and should be treated as preliminary. In particular, anything that appears to include financial or legal advice or proposed medical treatments should be treated with due caution. will not be responsible for any consequences of actions that result from any form of use of any documents on this website.

Add your own feedback and questions here:
You are equally welcome to be positive or negative about any paper but please be polite. If you are being critical you must mention at least one specific error, otherwise your comment will be deleted as unhelpful.

comments powered by Disqus