Virtual Face Animation Generation Based on Conditional Generative Adversarial Networks

Face animation is to generate facial animation video with rich expression according to the target face image. Virtual face animation generation has always been a hot topic in the field of computer vision. How to generate real and natural virtual face animation has always been a key issue in this field. In this paper, facial feature points are used to describe different facial shapes, which are used as guidance conditions for generating target facial animations. By improving the Conditional Generation Adversarial Network(CGAN), high quality virtual facial animations are generated. At the same time, the MOS evaluation verifies the effectiveness of the model in generating characters by virtual face animation.