A Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection

Advances in moving object detection have been driven by the active application of deep learning methods. However, many existing models render superior detection accuracy at the cost of high computational complexity and slow inference speed. This fact has hindered the development of such models in mo...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Bingxin Hou, Ying Liu, Nam Ling, Lingzhi Liu, Yongxiong Ren
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Convolutional neural network depthwise convolution moving object detection multi-input multi-output pointwise convolution scene independent evaluation Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/95b6bf8fc40a4a729d71a7693329d1e2
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:95b6bf8fc40a4a729d71a7693329d1e2
record_format	dspace
spelling	oai:doaj.org-article:95b6bf8fc40a4a729d71a7693329d1e22021-11-18T00:08:09ZA Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection2169-353610.1109/ACCESS.2021.3123975https://doaj.org/article/95b6bf8fc40a4a729d71a7693329d1e22021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9592757/https://doaj.org/toc/2169-3536Advances in moving object detection have been driven by the active application of deep learning methods. However, many existing models render superior detection accuracy at the cost of high computational complexity and slow inference speed. This fact has hindered the development of such models in mobile and embedded vision tasks, which need to be carried out in a timely fashion on a computationally limited platform. In this paper, we propose a super-fast (inference speed-154 fps) and lightweight (model size-1.45 MB) end-to-end 3D separable convolutional neural network with a multi-input multi-output (MIMO) strategy named “3DS_MM” for moving object detection. To improve detection accuracy, the proposed model adopts 3D convolution which is more suitable to extract both spatial and temporal information in video data than 2D convolution. To reduce model size and computational complexity, the standard 3D convolution is decomposed into depthwise and pointwise convolutions. Besides, we proposed a MIMO strategy to increase inference speed, which can take multiple frames as the network input and output multiple frames of detection results. Further, we conducted the scene dependent evaluation (SDE) and scene independent evaluation (SIE) on the benchmark CDnet2014 and DAVIS2016 datasets. Compared to state-of-the-art approaches, our proposed method significantly increases the inference speed, reduces the model size, meanwhile achieving the highest detection accuracy in the SDE setup and maintaining a competitive detection accuracy in the SIE setup.Bingxin HouYing LiuNam LingLingzhi LiuYongxiong RenIEEEarticleConvolutional neural networkdepthwise convolutionmoving object detectionmulti-input multi-outputpointwise convolutionscene independent evaluationElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 148433-148448 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Convolutional neural network depthwise convolution moving object detection multi-input multi-output pointwise convolution scene independent evaluation Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	Convolutional neural network depthwise convolution moving object detection multi-input multi-output pointwise convolution scene independent evaluation Electrical engineering. Electronics. Nuclear engineering TK1-9971 Bingxin Hou Ying Liu Nam Ling Lingzhi Liu Yongxiong Ren A Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection
description	Advances in moving object detection have been driven by the active application of deep learning methods. However, many existing models render superior detection accuracy at the cost of high computational complexity and slow inference speed. This fact has hindered the development of such models in mobile and embedded vision tasks, which need to be carried out in a timely fashion on a computationally limited platform. In this paper, we propose a super-fast (inference speed-154 fps) and lightweight (model size-1.45 MB) end-to-end 3D separable convolutional neural network with a multi-input multi-output (MIMO) strategy named “3DS_MM” for moving object detection. To improve detection accuracy, the proposed model adopts 3D convolution which is more suitable to extract both spatial and temporal information in video data than 2D convolution. To reduce model size and computational complexity, the standard 3D convolution is decomposed into depthwise and pointwise convolutions. Besides, we proposed a MIMO strategy to increase inference speed, which can take multiple frames as the network input and output multiple frames of detection results. Further, we conducted the scene dependent evaluation (SDE) and scene independent evaluation (SIE) on the benchmark CDnet2014 and DAVIS2016 datasets. Compared to state-of-the-art approaches, our proposed method significantly increases the inference speed, reduces the model size, meanwhile achieving the highest detection accuracy in the SDE setup and maintaining a competitive detection accuracy in the SIE setup.
format	article
author	Bingxin Hou Ying Liu Nam Ling Lingzhi Liu Yongxiong Ren
author_facet	Bingxin Hou Ying Liu Nam Ling Lingzhi Liu Yongxiong Ren
author_sort	Bingxin Hou
title	A Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection
title_short	A Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection
title_full	A Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection
title_fullStr	A Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection
title_full_unstemmed	A Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection
title_sort	fast lightweight 3d separable convolutional neural network with multi-input multi-output for moving object detection
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/95b6bf8fc40a4a729d71a7693329d1e2
work_keys_str_mv	AT bingxinhou afastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT yingliu afastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT namling afastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT lingzhiliu afastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT yongxiongren afastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT bingxinhou fastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT yingliu fastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT namling fastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT lingzhiliu fastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection AT yongxiongren fastlightweight3dseparableconvolutionalneuralnetworkwithmultiinputmultioutputformovingobjectdetection
_version_	1718425239478599680

A Fast Lightweight 3D Separable Convolutional Neural Network With Multi-Input Multi-Output for Moving Object Detection

Ejemplares similares