A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions

Object detection is extremely important in autonomous driving environment awareness. Besides vehicle and pedestrian detection, traffic signs and lights are important objects. The paper presents how to achieve precise results in multi-traffic object detection while minimizing the model size. A deep l...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Guoqiang Chen, Yanan Cheng
Formato:	article
Lenguaje:	EN
Publicado:	Tamkang University Press 2021
Materias:	ghost module attention mechanism dw convolution real-time object detection lightweight network complex traffic conditions Engineering (General). Civil engineering (General) TA1-2040 Chemical engineering TP155-156 Physics QC1-999
Acceso en línea:	https://doaj.org/article/457599980d104da9a852c81d3e6f55d5
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:457599980d104da9a852c81d3e6f55d5
record_format	dspace
spelling	oai:doaj.org-article:457599980d104da9a852c81d3e6f55d52021-11-27T11:08:36ZA lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions10.6180/jase.202206_25(3).00192708-99672708-9975https://doaj.org/article/457599980d104da9a852c81d3e6f55d52021-11-01T00:00:00Zhttp://jase.tku.edu.tw/articles/jase-202206-25-3-0019https://doaj.org/toc/2708-9967https://doaj.org/toc/2708-9975Object detection is extremely important in autonomous driving environment awareness. Besides vehicle and pedestrian detection, traffic signs and lights are important objects. The paper presents how to achieve precise results in multi-traffic object detection while minimizing the model size. A deep learning network YOLOv5s-Ghost-SE-DW is proposed based on the YOLOv5s. The proposed network can detect all traffic objects including traffic signs and lights. First, the convolution layer is replaced by the Ghost module to reduce the parameter and model size. Second, in order to improve accuracy and real-time performance, the attention mechanism SELayer is embedded to fuse more spatial features. Third, the DW convolution is used to extract features and further reduce the parameter number. The effect of different modules on the whole network is verified by ablation experiments. The YOLOv5s-Ghost-SE-DW yields a model size of 5.22MB while achieving 15.58 FPS real-time performance on the CPU. The FPS increases by 27.5%.Guoqiang ChenYanan ChengTamkang University Pressarticleghost moduleattention mechanismdw convolutionreal-time object detectionlightweight networkcomplex traffic conditionsEngineering (General). Civil engineering (General)TA1-2040Chemical engineeringTP155-156PhysicsQC1-999ENJournal of Applied Science and Engineering, Vol 25, Iss 3, Pp 527-535 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	ghost module attention mechanism dw convolution real-time object detection lightweight network complex traffic conditions Engineering (General). Civil engineering (General) TA1-2040 Chemical engineering TP155-156 Physics QC1-999
spellingShingle	ghost module attention mechanism dw convolution real-time object detection lightweight network complex traffic conditions Engineering (General). Civil engineering (General) TA1-2040 Chemical engineering TP155-156 Physics QC1-999 Guoqiang Chen Yanan Cheng A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions
description	Object detection is extremely important in autonomous driving environment awareness. Besides vehicle and pedestrian detection, traffic signs and lights are important objects. The paper presents how to achieve precise results in multi-traffic object detection while minimizing the model size. A deep learning network YOLOv5s-Ghost-SE-DW is proposed based on the YOLOv5s. The proposed network can detect all traffic objects including traffic signs and lights. First, the convolution layer is replaced by the Ghost module to reduce the parameter and model size. Second, in order to improve accuracy and real-time performance, the attention mechanism SELayer is embedded to fuse more spatial features. Third, the DW convolution is used to extract features and further reduce the parameter number. The effect of different modules on the whole network is verified by ablation experiments. The YOLOv5s-Ghost-SE-DW yields a model size of 5.22MB while achieving 15.58 FPS real-time performance on the CPU. The FPS increases by 27.5%.
format	article
author	Guoqiang Chen Yanan Cheng
author_facet	Guoqiang Chen Yanan Cheng
author_sort	Guoqiang Chen
title	A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions
title_short	A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions
title_full	A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions
title_fullStr	A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions
title_full_unstemmed	A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions
title_sort	lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions
publisher	Tamkang University Press
publishDate	2021
url	https://doaj.org/article/457599980d104da9a852c81d3e6f55d5
work_keys_str_mv	AT guoqiangchen alightweightmodelformultitrafficobjectdetectionbasedondeeplearningundercomplextrafficconditions AT yanancheng alightweightmodelformultitrafficobjectdetectionbasedondeeplearningundercomplextrafficconditions AT guoqiangchen lightweightmodelformultitrafficobjectdetectionbasedondeeplearningundercomplextrafficconditions AT yanancheng lightweightmodelformultitrafficobjectdetectionbasedondeeplearningundercomplextrafficconditions
_version_	1718408982886875136

A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions

Ejemplares similares