Multi-Branch Neural Architecture Search for Lightweight Image Super-Resolution

Deep convolutional neural networks (CNNs) are widely used to improve the performance of image restoration tasks, including single-image super-resolution (SISR). Generally, researchers are manually designing more complex and deeper CNNs to further increase the given problems’ performance....

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Joon Young Ahn, Nam Ik Cho
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Single image super-resolution neural architecture search image restoration Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/b28a0da03b2a4fa6ba2137eaed595b9b
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:b28a0da03b2a4fa6ba2137eaed595b9b
record_format	dspace
spelling	oai:doaj.org-article:b28a0da03b2a4fa6ba2137eaed595b9b2021-11-24T00:02:58ZMulti-Branch Neural Architecture Search for Lightweight Image Super-Resolution2169-353610.1109/ACCESS.2021.3127437https://doaj.org/article/b28a0da03b2a4fa6ba2137eaed595b9b2021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9612202/https://doaj.org/toc/2169-3536Deep convolutional neural networks (CNNs) are widely used to improve the performance of image restoration tasks, including single-image super-resolution (SISR). Generally, researchers are manually designing more complex and deeper CNNs to further increase the given problems’ performance. Instead of this hand-crafted CNN architecture design, neural architecture search (NAS) methods have been developed to find an optimal architecture for a given task automatically. For example, NAS-based SR methods find optimized network connections and operations by reinforcement learning (RL) or evolutionary algorithms (EA). These methods enable finding an optimal system automatically, but most of them need a very long search time. In this paper, we propose a new search method for the SISR that can significantly reduce the overall design time by applying a weight-sharing scheme. We also employ a multi-branch structure to enlarge the search space for capturing multi-scale features, resulting in better reconstruction on the textured region. Experiments show that the proposed method finds an optimal SISR network about twenty times faster than the existing methods, while showing comparable performance in terms of PSNR vs. parameters. Comparison of visual quality validates that the obtained SISR network reconstructs texture areas better than the previous methods because of the enlarged search space to find multi-scale features.Joon Young AhnNam Ik ChoIEEEarticleSingle image super-resolutionneural architecture searchimage restorationElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 153633-153646 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Single image super-resolution neural architecture search image restoration Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	Single image super-resolution neural architecture search image restoration Electrical engineering. Electronics. Nuclear engineering TK1-9971 Joon Young Ahn Nam Ik Cho Multi-Branch Neural Architecture Search for Lightweight Image Super-Resolution
description	Deep convolutional neural networks (CNNs) are widely used to improve the performance of image restoration tasks, including single-image super-resolution (SISR). Generally, researchers are manually designing more complex and deeper CNNs to further increase the given problems’ performance. Instead of this hand-crafted CNN architecture design, neural architecture search (NAS) methods have been developed to find an optimal architecture for a given task automatically. For example, NAS-based SR methods find optimized network connections and operations by reinforcement learning (RL) or evolutionary algorithms (EA). These methods enable finding an optimal system automatically, but most of them need a very long search time. In this paper, we propose a new search method for the SISR that can significantly reduce the overall design time by applying a weight-sharing scheme. We also employ a multi-branch structure to enlarge the search space for capturing multi-scale features, resulting in better reconstruction on the textured region. Experiments show that the proposed method finds an optimal SISR network about twenty times faster than the existing methods, while showing comparable performance in terms of PSNR vs. parameters. Comparison of visual quality validates that the obtained SISR network reconstructs texture areas better than the previous methods because of the enlarged search space to find multi-scale features.
format	article
author	Joon Young Ahn Nam Ik Cho
author_facet	Joon Young Ahn Nam Ik Cho
author_sort	Joon Young Ahn
title	Multi-Branch Neural Architecture Search for Lightweight Image Super-Resolution
title_short	Multi-Branch Neural Architecture Search for Lightweight Image Super-Resolution
title_full	Multi-Branch Neural Architecture Search for Lightweight Image Super-Resolution
title_fullStr	Multi-Branch Neural Architecture Search for Lightweight Image Super-Resolution
title_full_unstemmed	Multi-Branch Neural Architecture Search for Lightweight Image Super-Resolution
title_sort	multi-branch neural architecture search for lightweight image super-resolution
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/b28a0da03b2a4fa6ba2137eaed595b9b
work_keys_str_mv	AT joonyoungahn multibranchneuralarchitecturesearchforlightweightimagesuperresolution AT namikcho multibranchneuralarchitecturesearchforlightweightimagesuperresolution
_version_	1718416115392053248

Multi-Branch Neural Architecture Search for Lightweight Image Super-Resolution

Ejemplares similares