A Modified Generator for Multi-domain Transfer
Geum Boon Lee · 2020
We propose a modified generator for multi-domain transfer. It handles various data distributions and transfer the attributes of the data, while enabling high-quality image generation. We support the style translation for multi-domain and the scale-up style image generation. In addition, it works well with data sets other than face data. In order to process data with large statistical differences among domains, two factors are introduced. The results show that style transfer was well trained.