- 1
- 0
- 约4.32万字
- 约 8页
- 2023-12-25 发布于福建
- 举报
1762023,59(14)ComputerEngineeringandApplications计算机工程与应用
改进YOLOv5s算法的安全帽佩戴检测研究
1,21,2
祁泽政,徐银霞
1.武汉工程大学计算机科学与工程学院,武汉430205
2.武汉工程大学智能机器人湖北省重点实验室,武汉430205
摘要:针对安全帽佩戴检测在背景复杂、密集小目标等场景下的漏检、误检问题,基于YOLOv5s提出了一种适用
于安全帽佩戴检测的改进算法。采用串联形式的混合池化优化空间金字塔池化(spatialpyramidpooling,SPP)模
块,增强特征提取与特征表达,降低误检率;增加浅层输出加强深层特征图中目标位置信息的表达,提高小目标安全
帽佩戴检测的性能;在切片模块中嵌入坐标注意力机制(coordinateattention,CA)扩大感受野,加强位置信息与安全
帽特征的关联程度,提高安全帽目标检测的准确率。实验结果表明,改进算法的mAP达到了93.50%,比YOLOv5s、
YOLOX算法分别提高了4.28、5.14个百分点,达到了复杂背景下检测小目标和密集目标的要求。
关键词:YOLOv5s;深度学习;目标检测;人工智能;安全帽识别
文献标志码:A中图分类号:TP391.41doi:10.3778/j.issn.1002-8331.2303-0284
ResearchonHelmetWearingDetectionofImprovedYOLOv5sAlgorithm
QIZezheng1,2,XUYinxia1,2
1.SchoolofComputerScienceandEngineering,WuhanInstituteofTechnology,Wuhan430205,China
2.HubeiKeyLaboratoryofIntelligentRobot,WuhanInstituteofTechnology,Wuhan430205,China
Abstract:Aimingattheproblemofmisseddetectionandfalsedetectioninscenariossuchascomplexbackgroundand
densesmalltargets,animprovedalgorithmsuitableforsafetyhelmetwearingdetectionisproposedbasedonYOLOv5s.
Firstly,thehybridpoolingintandemformisusedtooptimizethespatialpyramidpooling(SPP)moduletoenhancefea-
tureextractionandfeatureexpressionandreducethefalsedetectionrate.Then,theshallowoutputisaddedtostrengthen
theexpressionoftargetpositioninformationinthedeepfeaturemap,andimprovetheperformanceofsmalltargetsafety
helmetwearingdetection.Finally,thecoordinateattentionmechanism(CA)isembeddedintheslicemoduletoexpand
thereceptivefield,strengthenthecorrelationbetweenpositioninformationandsafetyhelmetfeatures,andimprovetheaccu-
racyofsafetyhelmettargetdetection.ExperimentalresultsshowthatthemAPoftheimprovedalgorithmrea
原创力文档

文档评论(0)