CVPR2021

FaceInpainter: High Fidelity Face Adaptation to Heterogeneous Domains

Jia Li, Zhaoyang Li, Jie Cao, Xingguang Song, Ran He

Abstract

Figure 1: Face swapping results on wild face images by means of FaceInpainter. Our identity-guided face inpainting framework can produce high fidelity results in heterogeneous domains, i.e., photograph or non-photorealistic styles, e.g., oil painting, 3D cartoons, pencil drawing and exaggerated drawing.