FRGAN: A Blind Face Restoration with Generative Adversarial Networks

Recent works based on deep learning and facial priors have performed well in superresolving severely degraded facial images. However, due to the limitation of illumination, pixels of the monitoring probe itself, focusing area, and human motion, the face image is usually blurred or even deformed. To...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Tongxin Wei, Qingbao Li, Zhifeng Chen, Jinjin Liu
Formato: article
Lenguaje:EN
Publicado: Hindawi Limited 2021
Materias:
Acceso en línea:https://doaj.org/article/7ebc1be710db4890a41fd2491088c870
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:7ebc1be710db4890a41fd2491088c870
record_format dspace
spelling oai:doaj.org-article:7ebc1be710db4890a41fd2491088c8702021-11-08T02:37:28ZFRGAN: A Blind Face Restoration with Generative Adversarial Networks1563-514710.1155/2021/2384435https://doaj.org/article/7ebc1be710db4890a41fd2491088c8702021-01-01T00:00:00Zhttp://dx.doi.org/10.1155/2021/2384435https://doaj.org/toc/1563-5147Recent works based on deep learning and facial priors have performed well in superresolving severely degraded facial images. However, due to the limitation of illumination, pixels of the monitoring probe itself, focusing area, and human motion, the face image is usually blurred or even deformed. To address this problem, we properly propose Face Restoration Generative Adversarial Networks to improve the resolution and restore the details of the blurred face. They include the Head Pose Estimation Network, Postural Transformer Network, and Face Generative Adversarial Networks. In this paper, we employ the following: (i) Swish-B activation function that is used in Face Generative Adversarial Networks to accelerate the convergence speed of the cross-entropy cost function, (ii) a special prejudgment monitor that is added to improve the accuracy of the discriminant, and (iii) the modified Postural Transformer Network that is used with 3D face reconstruction network to correct faces at different expression pose angles. Our method improves the resolution of face image and performs well in image restoration. We demonstrate how our method can produce high-quality faces, and it is superior to the most advanced methods on the reconstruction task of blind faces for in-the-wild images; especially, our 8 × SR SSIM and PSNR are, respectively, 0.078 and 1.16 higher than FSRNet in AFLW.Tongxin WeiQingbao LiZhifeng ChenJinjin LiuHindawi LimitedarticleEngineering (General). Civil engineering (General)TA1-2040MathematicsQA1-939ENMathematical Problems in Engineering, Vol 2021 (2021)
institution DOAJ
collection DOAJ
language EN
topic Engineering (General). Civil engineering (General)
TA1-2040
Mathematics
QA1-939
spellingShingle Engineering (General). Civil engineering (General)
TA1-2040
Mathematics
QA1-939
Tongxin Wei
Qingbao Li
Zhifeng Chen
Jinjin Liu
FRGAN: A Blind Face Restoration with Generative Adversarial Networks
description Recent works based on deep learning and facial priors have performed well in superresolving severely degraded facial images. However, due to the limitation of illumination, pixels of the monitoring probe itself, focusing area, and human motion, the face image is usually blurred or even deformed. To address this problem, we properly propose Face Restoration Generative Adversarial Networks to improve the resolution and restore the details of the blurred face. They include the Head Pose Estimation Network, Postural Transformer Network, and Face Generative Adversarial Networks. In this paper, we employ the following: (i) Swish-B activation function that is used in Face Generative Adversarial Networks to accelerate the convergence speed of the cross-entropy cost function, (ii) a special prejudgment monitor that is added to improve the accuracy of the discriminant, and (iii) the modified Postural Transformer Network that is used with 3D face reconstruction network to correct faces at different expression pose angles. Our method improves the resolution of face image and performs well in image restoration. We demonstrate how our method can produce high-quality faces, and it is superior to the most advanced methods on the reconstruction task of blind faces for in-the-wild images; especially, our 8 × SR SSIM and PSNR are, respectively, 0.078 and 1.16 higher than FSRNet in AFLW.
format article
author Tongxin Wei
Qingbao Li
Zhifeng Chen
Jinjin Liu
author_facet Tongxin Wei
Qingbao Li
Zhifeng Chen
Jinjin Liu
author_sort Tongxin Wei
title FRGAN: A Blind Face Restoration with Generative Adversarial Networks
title_short FRGAN: A Blind Face Restoration with Generative Adversarial Networks
title_full FRGAN: A Blind Face Restoration with Generative Adversarial Networks
title_fullStr FRGAN: A Blind Face Restoration with Generative Adversarial Networks
title_full_unstemmed FRGAN: A Blind Face Restoration with Generative Adversarial Networks
title_sort frgan: a blind face restoration with generative adversarial networks
publisher Hindawi Limited
publishDate 2021
url https://doaj.org/article/7ebc1be710db4890a41fd2491088c870
work_keys_str_mv AT tongxinwei frganablindfacerestorationwithgenerativeadversarialnetworks
AT qingbaoli frganablindfacerestorationwithgenerativeadversarialnetworks
AT zhifengchen frganablindfacerestorationwithgenerativeadversarialnetworks
AT jinjinliu frganablindfacerestorationwithgenerativeadversarialnetworks
_version_ 1718443014538395648