Summary
International Workshop on Smart Info-Media Systems in Asia
2022
Session Number:RS2
Session:
Number:RS2-5
A Note on Expression Changes of Face Images in the Extended StyleGAN Latent Space
Yudai Iimori, Taiki Tsutsumi, Tomoaki Kimura, Hiroyuki Tsuji,
pp.119-123
Publication Date:2022/9/15
Online ISSN:2188-5079
DOI:10.34385/proc.69.RS2-5
PDF download
Summary:
In recent years, there has been remarkable progress in image generation techniques using GANs. The recent achievement of StyleGAN, which explicitly constructs a structured latent space W, enables semantic image processing based on manipulation of latent vectors. In this paper, we examine the fundamentals of performing facial expression change in face images on the extended latent space of StyleGAN. As a specific example, we identified a smile vector that changes a straight face expression into a smile, and presented a method to realize the change of eye area and mouth area separately, and confirmed its effectiveness.