Paper ID: 2408.15374

CycleGAN with Better Cycles

Tongzhou Wang, Yihan Lin

CycleGAN provides a framework to train image-to-image translation with unpaired datasets using cycle consistency loss [4]. While results are great in many applications, the pixel level cycle consistency can potentially be problematic and causes unrealistic images in certain cases. In this project, we propose three simple modifications to cycle consistency, and show that such an approach achieves better results with fewer artifacts.

Submitted: Aug 27, 2024