StarGAN is a generative adversarial network capable of learning mappings among multiple domains. It has a unified modeling architecture that allows simultaneous training of multiple datasets and different domains within a single network.
StarGAN is a novel and scalable approach to perform image-to-image translation among multiple
domains using a single model. It can translate input images to any target domain. It can generate higher visual quality images compared to existing methods. It can perform facial attribute transfer and facial expression synthesis.
StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation
Yunjey Choi, Minje Choi, Munyoung Kim, Jung-Woo Ha, Sunghun Kim and Jaegul Choo