DE-GAN: Domain Embedded GAN for High Quality Face Image Inpainting

作者：

Highlights：

• DE-GAN incorporates an HVAE in the generator to embed three types of face domain information into latent variables as the guidance for face inpainting, which produces more natural faces.

• Different from existing multiple stage training methods that use prior information to complete image or face inpainting, our proposed method is end-to-end trainable.

• To the best of our knowledge, our work is the first on the evaluation of the large pose side face inpainting problem. Our inpainting method achieves the state-of-the-art visual quality and facial structures for inpainting under-pose variations.

摘要

•DE-GAN incorporates an HVAE in the generator to embed three types of face domain information into latent variables as the guidance for face inpainting, which produces more natural faces.•Different from existing multiple stage training methods that use prior information to complete image or face inpainting, our proposed method is end-to-end trainable.•To the best of our knowledge, our work is the first on the evaluation of the large pose side face inpainting problem. Our inpainting method achieves the state-of-the-art visual quality and facial structures for inpainting under-pose variations.

论文关键词：Face Inpainting,Domain Embedding,Adversarial Generative Model

论文评审过程：Received 14 June 2021, Revised 17 September 2021, Accepted 31 October 2021, Available online 17 November 2021, Version of Record 28 February 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108415