A deep learning method for video‐based action recognition

Abstract In this paper, a deep learning method for video‐based action recognition is proposed. On the one hand, boundary compensation on the basis of a deep neural network is performed to achieve action proposal. Boundary compensation considering non‐maximum suppression according to sliding window p...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Guanwen Zhang, Yukun Rao, Changhao Wang, Wei Zhou, Xiangyang Ji
Formato:	article
Lenguaje:	EN
Publicado:	Wiley 2021
Materias:	Photography TR1-1050 Computer software QA76.75-76.765
Acceso en línea:	https://doaj.org/article/da8a99b21cce4c59b135963dcc5b04de
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:da8a99b21cce4c59b135963dcc5b04de
record_format	dspace
spelling	oai:doaj.org-article:da8a99b21cce4c59b135963dcc5b04de2021-11-29T03:38:16ZA deep learning method for video‐based action recognition1751-96671751-965910.1049/ipr2.12303https://doaj.org/article/da8a99b21cce4c59b135963dcc5b04de2021-12-01T00:00:00Zhttps://doi.org/10.1049/ipr2.12303https://doaj.org/toc/1751-9659https://doaj.org/toc/1751-9667Abstract In this paper, a deep learning method for video‐based action recognition is proposed. On the one hand, boundary compensation on the basis of a deep neural network is performed to achieve action proposal. Boundary compensation considering non‐maximum suppression according to sliding window priority is applied to remove redundant windows. To accurately detect boundaries, a boundary compensation network is established with multiple networks to process different numbers of segments. On the other hand, action recognition based on the resultant action proposals is performed. To further utilise boundary compensation, three methods are introduced for key frame selection. Optical flow and RGB features are combined via a channel fusion to realise feature representation. A two‐stream network with a spatiotemporal structure is adopted for action recognition. The proposed method is evaluated on three public datasets. The experimental results demonstrate that the proposed method achieves a superior performance to that of state‐of‐the‐art methods.Guanwen ZhangYukun RaoChanghao WangWei ZhouXiangyang JiWileyarticlePhotographyTR1-1050Computer softwareQA76.75-76.765ENIET Image Processing, Vol 15, Iss 14, Pp 3498-3511 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Photography TR1-1050 Computer software QA76.75-76.765
spellingShingle	Photography TR1-1050 Computer software QA76.75-76.765 Guanwen Zhang Yukun Rao Changhao Wang Wei Zhou Xiangyang Ji A deep learning method for video‐based action recognition
description	Abstract In this paper, a deep learning method for video‐based action recognition is proposed. On the one hand, boundary compensation on the basis of a deep neural network is performed to achieve action proposal. Boundary compensation considering non‐maximum suppression according to sliding window priority is applied to remove redundant windows. To accurately detect boundaries, a boundary compensation network is established with multiple networks to process different numbers of segments. On the other hand, action recognition based on the resultant action proposals is performed. To further utilise boundary compensation, three methods are introduced for key frame selection. Optical flow and RGB features are combined via a channel fusion to realise feature representation. A two‐stream network with a spatiotemporal structure is adopted for action recognition. The proposed method is evaluated on three public datasets. The experimental results demonstrate that the proposed method achieves a superior performance to that of state‐of‐the‐art methods.
format	article
author	Guanwen Zhang Yukun Rao Changhao Wang Wei Zhou Xiangyang Ji
author_facet	Guanwen Zhang Yukun Rao Changhao Wang Wei Zhou Xiangyang Ji
author_sort	Guanwen Zhang
title	A deep learning method for video‐based action recognition
title_short	A deep learning method for video‐based action recognition
title_full	A deep learning method for video‐based action recognition
title_fullStr	A deep learning method for video‐based action recognition
title_full_unstemmed	A deep learning method for video‐based action recognition
title_sort	deep learning method for video‐based action recognition
publisher	Wiley
publishDate	2021
url	https://doaj.org/article/da8a99b21cce4c59b135963dcc5b04de
work_keys_str_mv	AT guanwenzhang adeeplearningmethodforvideobasedactionrecognition AT yukunrao adeeplearningmethodforvideobasedactionrecognition AT changhaowang adeeplearningmethodforvideobasedactionrecognition AT weizhou adeeplearningmethodforvideobasedactionrecognition AT xiangyangji adeeplearningmethodforvideobasedactionrecognition AT guanwenzhang deeplearningmethodforvideobasedactionrecognition AT yukunrao deeplearningmethodforvideobasedactionrecognition AT changhaowang deeplearningmethodforvideobasedactionrecognition AT weizhou deeplearningmethodforvideobasedactionrecognition AT xiangyangji deeplearningmethodforvideobasedactionrecognition
_version_	1718407676850864128

A deep learning method for video‐based action recognition

Ejemplares similares