Generative Face Completion
CVPR 2017

1Univerisity of California, Merced       2Adobe Research

In this paper, we propose an effective face completion algorithm using a deep generative model. Different from well-studied background completion, the face completion task is more challenging as it often requires to generate semantically new pixels for the missing key components (e.g., eyes and mouths) that contain large appearance variations. Unlike existing nonparametric algorithms that search for patches to synthesize, our algorithm directly generates contents for missing regions based on a neural network. The model is trained with a combination of a reconstruction loss, two adversarial losses and a semantic parsing loss, which ensures pixel faithfulness and local-global contents consistency. With extensive experimental results, we demonstrate qualitatively and quantitatively that our model is able to deal with a large area of missing pixels in arbitrary shapes and generate realistic face completion results.


Completion demo

User editing demo

    author = {Li, Yijun and Liu, Sifei and Yang, Jimei and Yang, Ming-Hsuan},
    title = {Generative Face Completion},
    booktitle = {IEEE Conference on Computer Vision and Pattern Recognition},
    year = {2017}