A Survey to Deep Facial Attribute Manipulation Methods Part 3
Baseline CycleGAN ICCV2017 deadline: 201703 StarGAN CVPR2018 deadline: 201711
ECCV2018 Deadline:201803
Introduction Generates a high-res face image for the low-res input that satisfies the given attributes. motivation 1) Handle unpaired training. (low/high-res and high-res attribute images may not necessarily align with each other) 2) Allow easy control of the appearance of the generated face via the input attributes.
Method
Method
Special Section of CVM 2018
ECCV2018 Deadline:201803
arXiv201705
arXiv201803
Introduction Condition + Transfer motivation Incapability of generating image by exemplars; Being unable to transfer multiple face attributes simultaneously Low quality of generated images, such as low-resolution or artifacts.
Summary 创新点 功能 方法 单属性-多属性; Condition-transfer; Fader Network; GANimation (ECCV 2018 Best Paper Mention) 方法 面向思路:解耦;残差; 面向功能 图片质量、条件性