StarGAN v2: Diverse Image Synthesis for Multiple Domains
This work addresses the need for more versatile image synthesis models in computer vision, though it is incremental as it builds on prior methods like StarGAN.
The paper tackled the problem of image-to-image translation by proposing StarGAN v2, a single framework that improves diversity and scalability across multiple domains, showing significantly better results on CelebA-HQ and a new animal faces dataset (AFHQ) in terms of visual quality, diversity, and scalability.
A good image-to-image translation model should learn a mapping between different visual domains while satisfying the following properties: 1) diversity of generated images and 2) scalability over multiple domains. Existing methods address either of the issues, having limited diversity or multiple models for all domains. We propose StarGAN v2, a single framework that tackles both and shows significantly improved results over the baselines. Experiments on CelebA-HQ and a new animal faces dataset (AFHQ) validate our superiority in terms of visual quality, diversity, and scalability. To better assess image-to-image translation models, we release AFHQ, high-quality animal faces with large inter- and intra-domain differences. The code, pretrained models, and dataset can be found at https://github.com/clovaai/stargan-v2.