Generative Networks
摘要
Synthesizing photorealistic human faces is an appealing yet challenging task. Before the advent of deep learning, researchers used predefined 3D face models to design generative models for facial images. However, the abstraction and distortion of predefined models hinder the realism of the generated faces. With the development of deep learning, a large number of generative models have been proposed, especially in the field of face image generation.