A Style-Based Generator Architecture for Generative Adversarial Networks
125 بار بازدید -
6 سال پیش
-
Abstract:
We propose an alternative
Abstract:
We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis
Paper (PDF):
http://stylegan.xyz/paper
6 سال پیش
در تاریخ 1397/09/24 منتشر شده
است.
125
بـار بازدید شده