Article(id=1149776903822864434, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1149776900194791454, articleNumber=null, orderNo=null, doi=10.12404/j.issn.1671-1815.2403574, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=research-article, receivedDate=1715702400000, receivedDateStr=2024-05-15, revisedDate=1723219200000, revisedDateStr=2024-08-10, acceptedDate=null, acceptedDateStr=null, onlineDate=1752057775692, onlineDateStr=2025-07-09, pubDate=1744905600000, pubDateStr=2025-04-18, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1752057775692, onlineIssueDateStr=2025-07-09, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1752057775692, creator=13701087609, updateTime=1752057775692, updator=13701087609, issue=Issue{id=1149776900194791454, tenantId=1146029695717560320, journalId=1146123166801305609, year='2025', volume='25', issue='11', pageStart='4397', pageEnd='4826', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=0, createTime=1752057774827, creator=13701087609, updateTime=1768456666677, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1218558837930512931, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1149776900194791454, language=EN, specialIssueTitle=, coverIllustrator=, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1218558837930512932, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1149776900194791454, language=CN, specialIssueTitle=, coverIllustrator=, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=4666, endPage=4672, ext={EN=ArticleExt(id=1149776904099688500, articleId=1149776903822864434, tenantId=1146029695717560320, journalId=1146123166801305609, language=EN, title=Deep Learning Based Algorithm and Experiment for Ceramic Small Target Defect Detection, columnId=1156262729162810294, journalTitle=Science Technology and Engineering, columnName=Papers·Automation and Computational Technology, runingTitle=null, highlight=null, articleAbstract=

Defect detection is regarded as an indispensable step in the industrial production process. At present, manual detection is faced with the problems of low efficiency and high cost. A ceramic small target defect detection algorithm based on deep learning was proposed. For small target defects, a slice pre-training layer was first added to reduce the loss of graphics memory resources by large-size images. Secondly, a small target detection layer was added for the detection of small target defects, and a large target detection layer was removed to reduce the number of parameters. In addition, a feature selection fusion module based on MLCA (mixed local channel attention) was proposed to improve the perception of small target defects. Finally, a detection head with shared parameters was designed to further reduce the number of learnable parameters of the algorithm. By comparing with the baseline model, taking the ceramic cup as an example, the detection accuracy of this algorithm has been improved by 20.9%. Combined with the developed detection software and experimental platform, the detection efficiency of the ceramic cup has been enhanced by about 46.9%.

, correspAuthors=Xian-hai YANG, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Wen-jie MAO, Shi-long XIE, Lin-yu-xuan LI, Xian-hai YANG), CN=ArticleExt(id=1149776933883442099, articleId=1149776903822864434, tenantId=1146029695717560320, journalId=1146123166801305609, language=CN, title=基于深度学习的陶瓷小目标缺陷检测算法及实验, columnId=1156262729783567290, journalTitle=科学技术与工程, columnName=论文·自动化技术、计算机技术, runingTitle=null, highlight=null, articleAbstract=

缺陷检测在工业生产过程中是不可或缺的一步,目前人工检测存在效率低及成本高的问题,提出了一种基于深度学习的陶瓷小目标缺陷检测算法。针对小目标缺陷,本文算法首先添加切片预训练层,降低大尺寸图像对显卡内存资源的损耗;其次为小目标缺陷的检测添加小目标检测层,并去除大目标检测层,以减少参数量;另外提出一种基于MLCA(mixed local channel attention)的特征选择融合模块,提高对小目标缺陷的感知能力;最后设计了一种共享参数的检测头,进一步降低算法的可学习参数数量。通过与基线模型对比,以陶瓷杯为例,本文算法的检测精度提升了20.9%,结合研制的检测软件及实验平台,陶瓷杯检测效率提升了46.9%。

, correspAuthors=杨先海, authorNote=null, correspAuthorsNote=
* 杨先海(1963—),男,汉族,山东邹平人,教授,博士研究生导师。研究方向:机器视觉、机械创新设计。E-mail:
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=WSAHwm6jZIt0os0n3YLMvA==, magXml=esIs0dcZ04ZzFx1o54hABw==, pdfUrl=null, pdf=WMx1ZVDoJzFdzsxOK+dZsg==, pdfFileSize=6130104, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=3L+TQTXDkxTR5WkJmKTntw==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=qg8gS/bLtaavzOk1SQXd+Q==, mapNumber=null, authorCompany=null, fund=null, authors=

毛文杰(1999—),男,汉族,山东滨州人,硕士研究生。研究方向:机器视觉、机械创新设计。E-mail:

, authorsList=毛文杰, 谢世龙, 李林彧璇, 杨先海)}, authors=[Author(id=1218843914799337917, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=wenjie25v98@163.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1218843914891612612, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, authorId=1218843914799337917, language=EN, stringName=Wen-jie MAO, firstName=Wen-jie, middleName=null, lastName=MAO, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218843914996470219, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, authorId=1218843914799337917, language=CN, stringName=毛文杰, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=山东理工大学机械工程学院, 淄博 255049, bio={"content":"

毛文杰(1999—),男,汉族,山东滨州人,硕士研究生。研究方向:机器视觉、机械创新设计。E-mail:

"}, bioImg=null, bioContent=

毛文杰(1999—),男,汉族,山东滨州人,硕士研究生。研究方向:机器视觉、机械创新设计。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218843914690286002, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, xref=null, ext=[AuthorCompanyExt(id=1218843914698674609, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China), AuthorCompanyExt(id=1218843914702868914, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=山东理工大学机械工程学院, 淄博 255049)])]), Author(id=1218843915130687956, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1218843915256517087, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, authorId=1218843915130687956, language=EN, stringName=Shi-long XIE, firstName=Shi-long, middleName=null, lastName=XIE, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218843915445260773, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, authorId=1218843915130687956, language=CN, stringName=谢世龙, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=山东理工大学机械工程学院, 淄博 255049, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218843914690286002, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, xref=null, ext=[AuthorCompanyExt(id=1218843914698674609, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China), AuthorCompanyExt(id=1218843914702868914, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=山东理工大学机械工程学院, 淄博 255049)])]), Author(id=1218843915558506990, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1218843915713696246, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, authorId=1218843915558506990, language=EN, stringName=Lin-yu-xuan LI, firstName=Lin-yu-xuan, middleName=null, lastName=LI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218843915831136762, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, authorId=1218843915558506990, language=CN, stringName=李林彧璇, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=山东理工大学机械工程学院, 淄博 255049, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218843914690286002, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, xref=null, ext=[AuthorCompanyExt(id=1218843914698674609, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China), AuthorCompanyExt(id=1218843914702868914, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=山东理工大学机械工程学院, 淄博 255049)])]), Author(id=1218843915956965893, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=yxh@sdut.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1218843916078600714, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, authorId=1218843915956965893, language=EN, stringName=Xian-hai YANG, firstName=Xian-hai, middleName=null, lastName=YANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=*, address=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218843916170875407, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, authorId=1218843915956965893, language=CN, stringName=杨先海, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=*, address=山东理工大学机械工程学院, 淄博 255049, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218843914690286002, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, xref=null, ext=[AuthorCompanyExt(id=1218843914698674609, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China), AuthorCompanyExt(id=1218843914702868914, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=山东理工大学机械工程学院, 淄博 255049)])])], keywords=[Keyword(id=1218843916414145054, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, orderNo=1, keyword=defect detection), Keyword(id=1218843916527391268, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, orderNo=2, keyword=deep learning), Keyword(id=1218843916653220395, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, orderNo=3, keyword=small target defects), Keyword(id=1218843916820992564, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, orderNo=4, keyword=ceramic cups), Keyword(id=1218843916917461564, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, orderNo=1, keyword=缺陷检测), Keyword(id=1218843917009736260, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, orderNo=2, keyword=深度学习), Keyword(id=1218843917135565387, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, orderNo=3, keyword=小目标缺陷), Keyword(id=1218843917227840077, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, orderNo=4, keyword=陶瓷杯)], refs=[Reference(id=1218843920709112568, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=28, issue=9, pageStart=2587, pageEnd=2615, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=潘晓英, 贾凝心, 穆元震, journalName=中国图象图形学报, refType=null, unstructuredReference=潘晓英, 贾凝心, 穆元震, 等. 小目标检测研究综述[J]. 中国图象图形学报, 2023, 28(9): 2587-2615., articleTitle=小目标检测研究综述, refAbstract=null), Reference(id=1218843920834941690, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=28, issue=9, pageStart=2587, pageEnd=2615, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=Pan Xiaoying, Jia Ningxin, Mu Yuanzhen, journalName=Journal of Image and Graphics, refType=null, unstructuredReference=Pan Xiaoying, Jia Ningxin, Mu Yuanzhen, et al. Survey of small object detection[J]. Journal of Image and Graphics, 2023, 28(9): 2587-2615., articleTitle=Survey of small object detection, refAbstract=null), Reference(id=1218843920948187901, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2021, volume=47, issue=3, pageStart=293, pageEnd=302, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=员娇娇, 胡永利, 孙艳丰, journalName=北京工业大学学报, refType=null, unstructuredReference=员娇娇, 胡永利, 孙艳丰, 等. 基于深度学习的小目标检测方法综述[J]. 北京工业大学学报, 2021, 47(3): 293-302., articleTitle=基于深度学习的小目标检测方法综述, refAbstract=null), Reference(id=1218843921048851200, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2021, volume=47, issue=3, pageStart=293, pageEnd=302, url=null, language=null, rfNumber=[2], rfOrder=3, authorNames=Yuan Jiaojiao, Hu Yongli, Sun Yanfeng, journalName=Journal of Beijing University of Technology, refType=null, unstructuredReference=Yuan Jiaojiao, Hu Yongli, Sun Yanfeng, et al. Survey of small object detection methods based on deep learning[J]. Journal of Beijing University of Technology, 2021, 47(3): 293-302., articleTitle=Survey of small object detection methods based on deep learning, refAbstract=null), Reference(id=1218843921111765762, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=23, issue=36, pageStart=15538, pageEnd=15544, url=null, language=null, rfNumber=[3], rfOrder=4, authorNames=庄建军, 石潇愉, journalName=科学技术与工程, refType=null, unstructuredReference=庄建军, 石潇愉. 一种基于深度学习的商标检测方法[J]. 科学技术与工程, 2023, 23(36): 15538-15544., articleTitle=一种基于深度学习的商标检测方法, refAbstract=null), Reference(id=1218843921187263236, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=23, issue=36, pageStart=15538, pageEnd=15544, url=null, language=null, rfNumber=[3], rfOrder=5, authorNames=Zhuang Jianjun, Shi Xiaoyu, journalName=Science Technology and Engineering, refType=null, unstructuredReference=Zhuang Jianjun, Shi Xiaoyu. A trademark detection method based on deep learning[J]. Science Technology and Engineering, 2023, 23(36): 15538-15544., articleTitle=A trademark detection method based on deep learning, refAbstract=null), Reference(id=1218843921266955014, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2024, volume=224, issue=null, pageStart=113914, pageEnd=null, url=null, language=null, rfNumber=[4], rfOrder=6, authorNames=Cao T, Song K, Xu L, journalName=Measurement, refType=null, unstructuredReference=Cao T, Song K, Xu L, et al. Balanced multi-scale target score network for ceramic tile surface defect detection[J]. Measurement, 2024, 224: 113914., articleTitle=Balanced multi-scale target score network for ceramic tile surface defect detection, refAbstract=null), Reference(id=1218843921388589834, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=11, pageStart=120, pageEnd=128, url=null, language=null, rfNumber=[5], rfOrder=7, authorNames=邓钢, 赵庆华, 祁光威, journalName=现代制造工程, refType=null, unstructuredReference=邓钢, 赵庆华, 祁光威, 等. 基于YOLOv5的铝合金型材表面缺陷检测方法研究[J]. 现代制造工程, 2023(11): 120-128., articleTitle=基于YOLOv5的铝合金型材表面缺陷检测方法研究, refAbstract=null), Reference(id=1218843921514418960, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=11, pageStart=120, pageEnd=128, url=null, language=null, rfNumber=[5], rfOrder=8, authorNames=Deng Gang, Zhao Qinghua, Qi Guangwei, journalName=Modern Manufacturing Engineering, refType=null, unstructuredReference=Deng Gang, Zhao Qinghua, Qi Guangwei, et al. Research on aluminum profile surface defects detection method based on YOLOv5[J]. Modern Manufacturing Engineering, 2023(11): 120-128., articleTitle=Research on aluminum profile surface defects detection method based on YOLOv5, refAbstract=null), Reference(id=1218843921627665168, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=23, issue=27, pageStart=11730, pageEnd=11738, url=null, language=null, rfNumber=[6], rfOrder=9, authorNames=王宏, 韩晨, 袁伯阳, journalName=科学技术与工程, refType=null, unstructuredReference=王宏, 韩晨, 袁伯阳, 等. 基于改进YOLOv5的拥挤行人检测算法[J]. 科学技术与工程, 2023, 23(27): 11730-11738., articleTitle=基于改进YOLOv5的拥挤行人检测算法, refAbstract=null), Reference(id=1218843921728328466, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=23, issue=27, pageStart=11730, pageEnd=11738, url=null, language=null, rfNumber=[6], rfOrder=10, authorNames=Wang Hong, Han Chen, Yuan Boyang, journalName=Modern Manufacturing Engineering, refType=null, unstructuredReference=Wang Hong, Han Chen, Yuan Boyang, et al. Crowded pedestrian detection algorithm based on improved YOLOv5[J]. Modern Manufacturing Engineering, 2023, 23(27): 11730-11738., articleTitle=Crowded pedestrian detection algorithm based on improved YOLOv5, refAbstract=null), Reference(id=1218843921828991766, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2021, volume=21, issue=22, pageStart=9203, pageEnd=9211, url=null, language=null, rfNumber=[7], rfOrder=11, authorNames=洪腾蛟, 丁凤娟, 王鹏, journalName=科学技术与工程, refType=null, unstructuredReference=洪腾蛟, 丁凤娟, 王鹏, 等. 深度学习在轴承故障诊断领域的应用研究[J]. 科学技术与工程, 2021, 21(22): 9203-9211., articleTitle=深度学习在轴承故障诊断领域的应用研究, refAbstract=null), Reference(id=1218843921933849369, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2021, volume=21, issue=22, pageStart=9203, pageEnd=9211, url=null, language=null, rfNumber=[7], rfOrder=12, authorNames=Hong Tengjiao, Ding Fengjuan, Wang Peng, journalName=Modern Manufacturing Engineering, refType=null, unstructuredReference=Hong Tengjiao, Ding Fengjuan, Wang Peng, et al. Application of deep learning in bearing fault diagnosis[J]. Modern Manufacturing Engineering, 2021, 21(22): 9203-9211., articleTitle=Application of deep learning in bearing fault diagnosis, refAbstract=null), Reference(id=1218843922030318365, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2020, volume=22, issue=5, pageStart=420, pageEnd=427, url=null, language=null, rfNumber=[8], rfOrder=13, authorNames=金映谷, 张涛, 杨亚宁, journalName=大连民族大学学报, refType=null, unstructuredReference=金映谷, 张涛, 杨亚宁, 等. 基于深度学习的产品缺陷检测方法综述[J]. 大连民族大学学报, 2020, 22(5): 420-427., articleTitle=基于深度学习的产品缺陷检测方法综述, refAbstract=null), Reference(id=1218843922105815841, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2020, volume=22, issue=5, pageStart=420, pageEnd=427, url=null, language=null, rfNumber=[8], rfOrder=14, authorNames=Jin Yinggu, Zhang Tao, Yang Yaning, journalName=Journal of Dalian Minzu University, refType=null, unstructuredReference=Jin Yinggu, Zhang Tao, Yang Yaning, et al. Review of product defect detection methods based on deep learning[J]. Journal of Dalian Minzu University, 2020, 22(5): 420-427., articleTitle=Review of product defect detection methods based on deep learning, refAbstract=null), Reference(id=1218843922198090532, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=779, pageEnd=788, url=null, language=null, rfNumber=[9], rfOrder=15, authorNames=Edmon J, Divvala S, Girshick R, journalName=Conference on Computer Vision and Pattern Recognition (CVPR), refType=null, unstructuredReference=Edmon J, Divvala S, Girshick R, et al. You only look once: unified, real-time object detection[C]// Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE, 2016: 779-788., articleTitle=You only look once: unified, real-time object detection, refAbstract=null), Reference(id=1218843922290365224, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=21, pageEnd=37, url=null, language=null, rfNumber=[10], rfOrder=16, authorNames=Liu W, Anguelov D, Erhan D, journalName=Computer Vision-ECCV 2016, refType=null, unstructuredReference=Liu W, Anguelov D, Erhan D, et al. SSD: single shot multibox detector[C]// Computer Vision-ECCV 2016. Cham: Springer, 2016: 21-37., articleTitle=SSD: single shot multibox detector, refAbstract=null), Reference(id=1218843922403611436, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2017, volume=39, issue=6, pageStart=1137, pageEnd=1149, url=null, language=null, rfNumber=[11], rfOrder=17, authorNames=Ren S, He K, Girshick R, journalName=IEEE Transactions on Pattern Analysis & Machine Intelligence, refType=null, unstructuredReference=Ren S, He K, Girshick R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(6): 1137-1149., articleTitle=Faster R-CNN: towards real-time object detection with region proposal networks, refAbstract=null), Reference(id=1218843922483303213, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=51, issue=8, pageStart=2226, pageEnd=2237, url=null, language=null, rfNumber=[12], rfOrder=18, authorNames=王佳皓, 徐树公, 陆恒杰, journalName=电子学报, refType=null, unstructuredReference=王佳皓, 徐树公, 陆恒杰. 基于快速下采样的轻量化网络设计方法及人脸识别应用[J]. 电子学报, 2023, 51(8): 2226-2237., articleTitle=基于快速下采样的轻量化网络设计方法及人脸识别应用, refAbstract=null), Reference(id=1218843922562994991, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=51, issue=8, pageStart=2226, pageEnd=2237, url=null, language=null, rfNumber=[12], rfOrder=19, authorNames=Wang Jiahao, Xu Shugong, Lu Hengjie, journalName=Acta Electronica Sinica, refType=null, unstructuredReference=Wang Jiahao, Xu Shugong, Lu Hengjie. Lightweight network design and application for face recognition based on fast down-sampling[J]. Acta Electronica Sinica, 2023, 51(8): 2226-2237., articleTitle=Lightweight network design and application for face recognition based on fast down-sampling, refAbstract=null), Reference(id=1218843922734961459, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2021, volume=21, issue=31, pageStart=13430, pageEnd=13438, url=null, language=null, rfNumber=[13], rfOrder=20, authorNames=何永春, 申永伟, 吴涛, journalName=科学技术与工程, refType=null, unstructuredReference=何永春, 申永伟, 吴涛, 等. 基于注意力机制的多尺度仪表检测[J]. 科学技术与工程, 2021, 21(31): 13430-13438., articleTitle=基于注意力机制的多尺度仪表检测, refAbstract=null), Reference(id=1218843922823041845, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2021, volume=21, issue=31, pageStart=13430, pageEnd=13438, url=null, language=null, rfNumber=[13], rfOrder=21, authorNames=He Yongchun, Shen Yongwei, Wu Tao, journalName=Science Technology and Engineering, refType=null, unstructuredReference=He Yongchun, Shen Yongwei, Wu Tao, et al. Multi-scale instrument detection based on attention mechanism[J]. Science Technology and Engineering, 2021, 21(31): 13430-13438., articleTitle=Multi-scale instrument detection based on attention mechanism, refAbstract=null), Reference(id=1218843922906927927, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2023, volume=123, issue=null, pageStart=106442, pageEnd=null, url=null, language=null, rfNumber=[14], rfOrder=22, authorNames=Wan D, Lu R, Shen S, journalName=Engineering Applications of Artificial Intelligence, refType=null, unstructuredReference=Wan D, Lu R, Shen S, et al. Mixed local channel attention for object detection[J]. Engineering Applications of Artificial Intelligence, 2023, 123: 106442., articleTitle=Mixed local channel attention for object detection, refAbstract=null), Reference(id=1218843922982425402, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2024, volume=61, issue=8, pageStart=245, pageEnd=255, url=null, language=null, rfNumber=[15], rfOrder=23, authorNames=周颖, 颜毓泽, 陈海永, journalName=激光与光电子学进展, refType=null, unstructuredReference=周颖, 颜毓泽, 陈海永, 等. 基于改进YOLOv8的光伏电池缺陷检测[J]. 激光与光电子学进展, 2024, 61(8): 245-255., articleTitle=基于改进YOLOv8的光伏电池缺陷检测, refAbstract=null), Reference(id=1218843923078894396, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, doi=null, pmid=null, pmcid=null, year=2024, volume=61, issue=8, pageStart=245, pageEnd=255, url=null, language=null, rfNumber=[15], rfOrder=24, authorNames=Zhou Ying, Yan Shuze, Chen Haiyong, journalName=Laser & Optoelectronics Progress, refType=null, unstructuredReference=Zhou Ying, Yan Shuze, Chen Haiyong, et al. Defect detection of photovoltaic cells based on improved YOLOV8[J]. Laser & Optoelectronics Progress, 2024, 61(8): 245-255., articleTitle=Defect detection of photovoltaic cells based on improved YOLOV8, refAbstract=null)], funds=[Fund(id=1218843920499397364, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, awardId=52075306, language=CN, fundingSource=国家自然科学基金(52075306), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1218843914690286002, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, xref=null, ext=[AuthorCompanyExt(id=1218843914698674609, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China), AuthorCompanyExt(id=1218843914702868914, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, companyId=1218843914690286002, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=山东理工大学机械工程学院, 淄博 255049)])], figs=[ArticleFig(id=1218843917500469851, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.1, caption=Defect example, figureFileSmall=/KohT+tn+VlJzvvz9lXOyw==, figureFileBig=No08ENtjZSoShr6O/gFU4Q==, tableContent=null), ArticleFig(id=1218843917571773025, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图1, caption=缺陷示例, figureFileSmall=/KohT+tn+VlJzvvz9lXOyw==, figureFileBig=No08ENtjZSoShr6O/gFU4Q==, tableContent=null), ArticleFig(id=1218843917685019241, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.2, caption=Structure diagram of the YOLOv8 algorithm model, figureFileSmall=MlaVBwr+8i7V+ZjN/8oNFg==, figureFileBig=navtAYjVyXUdG6jggLRIBw==, tableContent=null), ArticleFig(id=1218843917789876850, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图2, caption=YOLOv8算法模型结构图, figureFileSmall=MlaVBwr+8i7V+ZjN/8oNFg==, figureFileBig=navtAYjVyXUdG6jggLRIBw==, tableContent=null), ArticleFig(id=1218843917877957242, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.3, caption=Image segmentation map, figureFileSmall=9GCaoL4CNyoNGsvEJAqDUQ==, figureFileBig=/HC9SkJNhXrciwbaEdirLg==, tableContent=null), ArticleFig(id=1218843917987009150, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图3, caption=图像分割图, figureFileSmall=9GCaoL4CNyoNGsvEJAqDUQ==, figureFileBig=/HC9SkJNhXrciwbaEdirLg==, tableContent=null), ArticleFig(id=1218843918108643970, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.4, caption=C2f structure diagram, figureFileSmall=ZjZxkC0VAzi6B6fCtRE1WQ==, figureFileBig=c7nbeJijV4d2Ps9qm2yjVA==, tableContent=null), ArticleFig(id=1218843918234473098, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图4, caption=C2f结构图, figureFileSmall=ZjZxkC0VAzi6B6fCtRE1WQ==, figureFileBig=c7nbeJijV4d2Ps9qm2yjVA==, tableContent=null), ArticleFig(id=1218843918343525005, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.5, caption=MLCA feature selection module structure diagram, figureFileSmall=+GXMR97VQqEo7zqavPgLrw==, figureFileBig=Uh4g7VpgcWZ7sp/eSY3kfw==, tableContent=null), ArticleFig(id=1218843918452576917, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图5, caption=MLCA特征选择模块结构图, figureFileSmall=+GXMR97VQqEo7zqavPgLrw==, figureFileBig=Uh4g7VpgcWZ7sp/eSY3kfw==, tableContent=null), ArticleFig(id=1218843918570017434, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.6, caption=MLC-C2f feature selection fusion module, figureFileSmall=UhEgSiqEdyVjgFJsdamYbQ==, figureFileBig=IU1KbfemNaoqoY7SUvDAWA==, tableContent=null), ArticleFig(id=1218843918641320612, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图6, caption=MLC-C2f特征选择融合模块, figureFileSmall=UhEgSiqEdyVjgFJsdamYbQ==, figureFileBig=IU1KbfemNaoqoY7SUvDAWA==, tableContent=null), ArticleFig(id=1218843918733595306, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.7, caption=YOLOv8 detection head structure, figureFileSmall=U96NIbwc68lbiiXKPEHKaw==, figureFileBig=ldXEqtoHGDsVbbkKgRB1CA==, tableContent=null), ArticleFig(id=1218843918867813041, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图7, caption=YOLOv8检测头结构, figureFileSmall=U96NIbwc68lbiiXKPEHKaw==, figureFileBig=ldXEqtoHGDsVbbkKgRB1CA==, tableContent=null), ArticleFig(id=1218843918997836471, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.8, caption=Shared parameter detection head, figureFileSmall=5jWPkji9A25RLSq0+Q1Piw==, figureFileBig=SQRCGV53jqtaPbzhV2WKbw==, tableContent=null), ArticleFig(id=1218843919136248508, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图8, caption=共享参数检测头, figureFileSmall=5jWPkji9A25RLSq0+Q1Piw==, figureFileBig=SQRCGV53jqtaPbzhV2WKbw==, tableContent=null), ArticleFig(id=1218843919232717508, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.9, caption=Structure diagram of ceramic cup small target defect detection algorithm based on YOLOv8, figureFileSmall=5wF/Tyuvq4k3MgQcIWpC0w==, figureFileBig=Hb94XWMomG842n2F2LyKzg==, tableContent=null), ArticleFig(id=1218843919366935241, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图9, caption=基于YOLOv8的陶瓷杯小目标缺陷检测算法结构图, figureFileSmall=5wF/Tyuvq4k3MgQcIWpC0w==, figureFileBig=Hb94XWMomG842n2F2LyKzg==, tableContent=null), ArticleFig(id=1218843919480181451, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.10, caption=Automatic detection of the experimental platform, figureFileSmall=hWUhCkKW44/whYWKVACjfw==, figureFileBig=JJzgfCxuAlM0XG9ngDk3Kw==, tableContent=null), ArticleFig(id=1218843919601816270, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图10, caption=自动检测实验平台

1为PC端;2为开关量采集模块;3为光电开关;4为工业相机;5为传送带;6为收集箱;7为UR5机器人

, figureFileSmall=hWUhCkKW44/whYWKVACjfw==, figureFileBig=JJzgfCxuAlM0XG9ngDk3Kw==, tableContent=null), ArticleFig(id=1218843919694090963, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Fig.11, caption=Algorithm detection time, figureFileSmall=Dze3nxoCmWlwcFEz2edIZA==, figureFileBig=h0Vmw9KmrITV4NiwpAK/VA==, tableContent=null), ArticleFig(id=1218843919765394135, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=图11, caption=算法检测时间, figureFileSmall=Dze3nxoCmWlwcFEz2edIZA==, figureFileBig=h0Vmw9KmrITV4NiwpAK/VA==, tableContent=null), ArticleFig(id=1218843919836697308, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Table 1, caption=

YOLOv8 detection target correspondence table

, figureFileSmall=null, figureFileBig=null, tableContent=
检测的特征图/
(pixel×pixel)
检测目标大小/
(pixel×pixel)
P3 80×80 8×8
P4 40×40 16×16
P5 20×20 32×32
), ArticleFig(id=1218843919937360609, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=表1, caption=

YOLOv8检测目标对应表

, figureFileSmall=null, figureFileBig=null, tableContent=
检测的特征图/
(pixel×pixel)
检测目标大小/
(pixel×pixel)
P3 80×80 8×8
P4 40×40 16×16
P5 20×20 32×32
), ArticleFig(id=1218843920046412518, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Table 2, caption=

Representation of confusion matrix

, figureFileSmall=null, figureFileBig=null, tableContent=
项目 预测类别
True False
实际类别 True TP (True Positive) FN (False Negative)
False FP (False Positive) TN (True Negative)
), ArticleFig(id=1218843920147075817, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=表2, caption=

混淆矩阵的表示

, figureFileSmall=null, figureFileBig=null, tableContent=
项目 预测类别
True False
实际类别 True TP (True Positive) FN (False Negative)
False FP (False Positive) TN (True Negative)
), ArticleFig(id=1218843920247739117, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=EN, label=Table 3, caption=

Comparative ablation experiment

, figureFileSmall=null, figureFileBig=null, tableContent=
名称 P R mAP/
%
参数量/
M
YOLOv8原尺寸图训练 0.813 0.740 76.6 3.0
YOLOv8+切片 0.942 0.932 95.5 3.0
YOLOv8+切片+重构 0.937 0.917 96.1 1.0
YOLOv8+切片+
重构+MLC-C2f
0.944 0.922 96.7 1.0
YOLOv8+切片+重构+
MLC-C2f+SP-Detect
0.974 0.899 97.5 0.99
), ArticleFig(id=1218843920327430895, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1149776903822864434, language=CN, label=表3, caption=

消融对比实验

, figureFileSmall=null, figureFileBig=null, tableContent=
名称 P R mAP/
%
参数量/
M
YOLOv8原尺寸图训练 0.813 0.740 76.6 3.0
YOLOv8+切片 0.942 0.932 95.5 3.0
YOLOv8+切片+重构 0.937 0.917 96.1 1.0
YOLOv8+切片+
重构+MLC-C2f
0.944 0.922 96.7 1.0
YOLOv8+切片+重构+
MLC-C2f+SP-Detect
0.974 0.899 97.5 0.99
)], attaches=null, journal=Journal(id=1146119176004939786, delFlag=0, nameCn=科学技术与工程, nameEn=Science Technology and Engineering, nameHistory1=null, nameHistory2=null, issn=1671-1815, eissn=, cn=11-4688/T, coden=null, periodic=4, language=CN, oaType=是, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=UKU/O7GSka5polgCTkbIIw==, journalPrice=null, startedYear=null, abbrevIsoEn=Sci Technol Eng, journalRemark=null, publicationField=null, createdTime=null, updatedTime=1754445529766, createdBy=null, updatedBy=13701087609, firstLetterCn=S, firstLetterEn=S, subjectCode=Natural Sciences, subjectName=自然科学, subjectCodeEn=Natural Sciences, subjectNameEn=null, picCn=UKU/O7GSka5polgCTkbIIw==, picEn=5hwlULoNwcbj3xUmVi9MAQ==, jcr=null, cjcr=null, exts=[JournalExt(id=1159791870395564357, language=CN, name=科学技术与工程, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.stae.com.cn/jsygc/home, createdTime=1754445529793, updatedTime=1754445529793, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=http://www.stae.com.cn/jsygc/site/menus/20090429150146001, submissionAuthorUrl=http://www.stae.com.cn/jsygc/author/login, submissionEditorUrl=http://www.stae.com.cn/jsygc/editor/login, submissionReviewUrl=http://www.stae.com.cn/jsygc/reviewer/login, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1159791870441701702, language=EN, name=Science Technology and Engineering, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.stae.com.cn/jsygc/home, createdTime=1754445529804, updatedTime=1754445529804, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.stae.com.cn/jsygc/author/login, submissionEditorUrl=http://www.stae.com.cn/jsygc/editor/login, submissionReviewUrl=http://www.stae.com.cn/jsygc/reviewer/login, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1146123166801305609, websiteList=[Website(id=1148243202391400884, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146123166801305609, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/kxjsygc/CN, language=CN, createTime=1751692112777, createBy=18614031015, updateTime=1753520965431, updateBy=18614031015, name=科学技术与工程-中文站点, tplId=1146099689490845704, title=科学技术与工程, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1148622798802673703, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=articleTextType, value=kx, createTime=1751782615614, updateTime=1751782615614, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798781702180, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=banner, value=null, createTime=1751782615609, updateTime=1751782615609, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798769119267, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=j86gbwi+p0Idkyl5SzIlmQ==, createTime=1751782615606, updateTime=1751782615606, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798794285094, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1751782615612, updateTime=1751782615612, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798790090789, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1751782615611, updateTime=1751782615611, creator=18614031015, updator=18614031015)]), Website(id=1155914124811976731, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146123166801305609, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/kxjsygc/EN, language=EN, createTime=1753521003206, createBy=18614031015, updateTime=1753521003206, updateBy=18614031015, name=科学技术与工程-英文站点, tplId=1146101810881728533, title=Science Technology and Engineering, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1155914371227308235, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=articleTextType, value=kx, createTime=1753521061952, updateTime=1753521061952, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371210531016, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=banner, value=null, createTime=1753521061947, updateTime=1753521061947, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371202142407, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=j86gbwi+p0Idkyl5SzIlmQ==, createTime=1753521061945, updateTime=1753521061945, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371223113930, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1753521061950, updateTime=1753521061950, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371218919625, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1753521061949, updateTime=1753521061949, creator=18614031015, updator=18614031015)])], journalTitle=科学技术与工程, weixinUrl=null, journalUrl=null, iacademicId=null, status=0, seqNo=null, journalTitleEn=Science Technology and Engineering, journalPhotoCn=UKU/O7GSka5polgCTkbIIw==, journalPhotoEn=5hwlULoNwcbj3xUmVi9MAQ==, journalFirstLetter=S, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=null, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/kxjsygc/CN/10.12404/j.issn.1671-1815.2403574, detailUrlEn=https://castjournals.cast.org.cn/joweb/kxjsygc/EN/10.12404/j.issn.1671-1815.2403574, pdfUrlCn=https://castjournals.cast.org.cn/joweb/kxjsygc/CN/PDF/10.12404/j.issn.1671-1815.2403574, pdfUrlEn=https://castjournals.cast.org.cn/joweb/kxjsygc/EN/PDF/10.12404/j.issn.1671-1815.2403574, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于深度学习的陶瓷小目标缺陷检测算法及实验
收藏切换
PDF下载
毛文杰 , 谢世龙 , 李林彧璇 , 杨先海 *
科学技术与工程 | 论文·自动化技术、计算机技术 2025,25(11): 4666-4672
收起
收藏切换
科学技术与工程 | 论文·自动化技术、计算机技术 2025, 25(11): 4666-4672
基于深度学习的陶瓷小目标缺陷检测算法及实验
全屏
毛文杰 , 谢世龙, 李林彧璇, 杨先海*
作者信息
  • 山东理工大学机械工程学院, 淄博 255049
  • 毛文杰(1999—),男,汉族,山东滨州人,硕士研究生。研究方向:机器视觉、机械创新设计。E-mail:

通讯作者:

* 杨先海(1963—),男,汉族,山东邹平人,教授,博士研究生导师。研究方向:机器视觉、机械创新设计。E-mail:
Deep Learning Based Algorithm and Experiment for Ceramic Small Target Defect Detection
Wen-jie MAO , Shi-long XIE, Lin-yu-xuan LI, Xian-hai YANG*
Affiliations
  • School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China
出版时间: 2025-04-18 doi: 10.12404/j.issn.1671-1815.2403574
文章导航
收藏切换

缺陷检测在工业生产过程中是不可或缺的一步,目前人工检测存在效率低及成本高的问题,提出了一种基于深度学习的陶瓷小目标缺陷检测算法。针对小目标缺陷,本文算法首先添加切片预训练层,降低大尺寸图像对显卡内存资源的损耗;其次为小目标缺陷的检测添加小目标检测层,并去除大目标检测层,以减少参数量;另外提出一种基于MLCA(mixed local channel attention)的特征选择融合模块,提高对小目标缺陷的感知能力;最后设计了一种共享参数的检测头,进一步降低算法的可学习参数数量。通过与基线模型对比,以陶瓷杯为例,本文算法的检测精度提升了20.9%,结合研制的检测软件及实验平台,陶瓷杯检测效率提升了46.9%。

缺陷检测  /  深度学习  /  小目标缺陷  /  陶瓷杯

Defect detection is regarded as an indispensable step in the industrial production process. At present, manual detection is faced with the problems of low efficiency and high cost. A ceramic small target defect detection algorithm based on deep learning was proposed. For small target defects, a slice pre-training layer was first added to reduce the loss of graphics memory resources by large-size images. Secondly, a small target detection layer was added for the detection of small target defects, and a large target detection layer was removed to reduce the number of parameters. In addition, a feature selection fusion module based on MLCA (mixed local channel attention) was proposed to improve the perception of small target defects. Finally, a detection head with shared parameters was designed to further reduce the number of learnable parameters of the algorithm. By comparing with the baseline model, taking the ceramic cup as an example, the detection accuracy of this algorithm has been improved by 20.9%. Combined with the developed detection software and experimental platform, the detection efficiency of the ceramic cup has been enhanced by about 46.9%.

defect detection  /  deep learning  /  small target defects  /  ceramic cups
毛文杰, 谢世龙, 李林彧璇, 杨先海. 基于深度学习的陶瓷小目标缺陷检测算法及实验. 科学技术与工程, 2025 , 25 (11) : 4666 -4672 . DOI: 10.12404/j.issn.1671-1815.2403574
Wen-jie MAO, Shi-long XIE, Lin-yu-xuan LI, Xian-hai YANG. Deep Learning Based Algorithm and Experiment for Ceramic Small Target Defect Detection[J]. Science Technology and Engineering, 2025 , 25 (11) : 4666 -4672 . DOI: 10.12404/j.issn.1671-1815.2403574
陶瓷材料因其独特的物理、化学和机械性能,在许多行业中发挥着重要作用。如在航空航天领域,先进的陶瓷材料可用于航空发动机的热障涂层,提高航空发动机的性能和寿命;在军工领域,陶瓷材料可用于火箭及导弹的雷达保护罩,抵御高速飞行状态中遇到的各种极端环境,同时对雷达信号的传输具有低损耗特性。然而,陶瓷材料在制造的过程中难免会出现一些缺陷,如黑点、裂纹、划痕等,这些缺陷虽小,却有可能对材料的最终应用造成重大影响,因此在陶瓷材料投入使用之前,对其进行严格的缺陷检测是至关重要的一环。
当前陶瓷制品的缺陷检测,主要依赖于人工视觉检查,这一方法存在若干局限性。首先,由于某些瑕疵尺寸微小,人工检测容易出现误判,这直接影响了检测的准确性。其次,人工检测的效率较低,尤其是在面对大批量产品时,难以满足工业生产的需求。此外,传统的检测通常在高强度光源下进行,长时间的视觉作业容易导致检测人员视觉疲劳,甚至可能对视力造成不可逆的损伤,给工作人员的身体健康带来严重威胁。
以陶瓷杯为例,对其缺陷进行检测,而在陶瓷杯众多的缺陷中,黑点缺陷属于最小的缺陷,在检测的过程中也最容易被忽视,黑点缺陷的直径约0.5 mm,严格属于小目标的范畴,小目标检测是目标检测领域中的一个重要方向,也是一个具有挑战性的问题[1]
在检测的过程中,小目标缺陷携带的特征信息是较少的,并缺乏丰富的空间上下文信息,同时也容易受到背景干扰影响,另外小目标缺陷的尺寸变化范围较大,需要检测模型有良好的尺度适应性,从而也使得检测的难度以及成本都大大增加[2]
在小目标检测方面,庄建军等[3]提出了一种基于深度学习的商标检测方法,其主要使用了多尺度特征融合网络SimBiFPN(sim bidirectional feature pyramid network)以及引入注意力机制等方法提升YOLOv7-tiny对小目标商标检测的准确率;Cao等[4]为解决瓷砖缺陷较小的问题,提出了一种平衡多尺度目标评分网络,其主要引入内容感知模块来提升YOLOv5的检测性能;邓钢等[5]通过将K-Means算法应用到自适应锚框算法中,提出一种基于YOLOv5铝合金型材表面缺陷检测方法,提升了检测的精度;王宏等[6]针对密集人群尺寸偏小而检测效果不佳的问题,通过在YOLOv5中添加坐标注意力机制以及添加浅层检测尺度等方法,提升了其检测的准确率。
上述研究取得了一定的研究成果,但陶瓷杯黑点缺陷更小,不易被检测,另外针对陶瓷杯的缺陷检测研究较少。因此针对这些问题,将对陶瓷杯上的黑点小目标缺陷做更进一步的研究,实现对陶瓷杯小目标缺陷快速、准确的检测。
深度学习引入了端到端学习的概念,即输入端与输出端直接进行连接,无需对中间层进行干预,从而实现目标检测[7]。与传统的机器视觉检测方法相比,此类方法能够更好地适应复杂多变的实际场景,提供更快速、更准确的检测效果[8]
在目标检测中常用的深度学习算法模型有YOLO(you only look once)[9]、SSD(single shot multibox detector)[10]、Faster R-CNN(faster region-based convolutional neural network)[11],在其中YOLO算法是比较流行的目标检测模型,最初于2015年提出,随着其不断地发展,新一代的YOLOv8模型参数量更少,速度更快,并且本次解决的小目标如图1所示,缺陷的尺寸足够小,而YOLOv8对小目标和少样本问题有很大帮助。因此,选用了YOLOv8算法并在此基础上进行改进优化。
YOLOv8是由骨干网络(Backbone)、颈部网络(Neck)、检测头网络(Head)组成,其中Backbone对输入图像进行特征提取,以捕捉图像中的上下文信息和关键特征;Neck用于不同尺度图像之间的特征融合,通过将来自不同层级的特征图进行有效融合,从而实现多尺度目标检测;Head负责执行最终的目标检测和回归预测,将提取到的特征转化为目标的位置和类别信息。这种结构的组合使得YOLOv8具有较强的目标检测和识别能力,能够在复杂场景下有效地实现目标定位和识别,其结构如图2所示。
由于所需检测的缺陷属于小目标缺陷,使得YOLOv8的训练和检测效果不佳,其准确率也达不到既定要求,因此对YOLOv8进行改进,提出了一种针对陶瓷杯小目标缺陷的深度学习检测算法。
为保证深度学习算法的推理速度,在开始训练前会执行resize操作,即将输入图片的大小缩放为640像素×640像素,以提高整体的运行速度,在YOLOv8中,为了减小信息的丢失使用letterbox方法(一种自适应图片缩放技术)进行缩放,以提高目标检测算法的性能,但由于黑点缺陷的尺寸过于小,大多数的缺陷信息依然被丢失,造成学习效果差。而使用原图(2 592像素×1 944像素,500万像素工业相机所采集)进行训练时,训练速度及检测精度都不理想,并且所消耗内存较高,不利于检测工作的进行,因此对其进行改进。
针对输入图像过大的问题,提出了一种切片训练预处理的方法,即在训练前将原图进行切片处理,并为了进一步减少运算量,提高训练速度,将不带标签的图像进行剔除,随后再进行训练,就陶瓷杯检测而言,为保证切割后的图像尽可能显示出黑点缺陷,故将其分割为20份,之后再进行训练,分割后的图像如图3所示。
通过这种方法,提高了YOLOv8的训练速度,降低了近一倍的显卡内存损耗,并提高了深度学习算法模型的特征学习能力,为陶瓷杯小目标缺陷的快速、准确检测打下基础。
下采样在深度学习扮演着至关重要的角色,通过下采样操作,可以有效地减少特征图的维度,保留主要的特征信息,并减少后续层的计算量,以提升检测模型的性能,这些都是对于一般检测物体而言[12]。但是对于小目标来说,多次的下采样操作会导致原始图像的空间信息丢失,使小目标更难以被有效识别,另外,随着下采样操作的进行,小目标会变得更小,较深的特征图很难学习到小目标缺陷的特征信息,并且深度学习模型的位置检测精度也大幅度降低。
为了提高YOLOv8对小目标缺陷的感知能力,故将下采样操作的层数减少,提高其特征学习能力。在YOLOv8原模型中,含有三个检测头,具体如表1所示。
由于检测的目标较小,P5层的检测头发挥的作用不明显,因此同样将其去除,使YOLOv8能够更加地关注到小目标信息,增强对小目标的敏感性,从而提升整体的检测精度。通过下采样层数以及大目标检测头的去除,大幅度降低了YOLOv8的参数数量,形成了一种轻量级的模型框架,为其在工业中的部署提供了便利。
C2f(cspdarknet53 to 2-stage feature pyramid network)是YOLOv8中的核心模块之一,其主要目的是将主干路径和分支路径的特征进行融合,以提高特征图的表达能力,通过这种方式,获得来自不同层级的特征图的丰富信息,从而提高网络的感知能力和检测性能。其结构如图4所示,从中可以看出其主要是由多层的Bottleneck模块构成,而由于黑点缺陷特征小的缘故,其特征提取效果较差,因此对其改进。
特征选择模块在各大神经网络中发挥出了明显的效果,其能够帮助模型更好地关注目标区域,以提高模型检测的准确性,同时它还可以减少图像中的背景噪声,继而提升模型的鲁棒性[13]。多数的特征选择模块只包含了通道信息,但是在小目标缺陷的检测中其空间信息也尤为重要。
MLCA (mixed local channel attention)是一种轻量级的特征选择模块[14]。其首先通过自适应平均池化计算局部平均值和全局平均值,然后使用一维卷积对局部平均值和全局平均值进行卷积操作,并对结果进行形状和转置操作。最后,通过应用sigmoid函数计算权重,并根据权重对输入进行加权,实现对目标的选择,其结构如图5所示。
在其中MLCA使用的是一维卷积,明显减少了计算量和参数量,其卷积核大小k与通道数正比,而k的计算公式为
k=Φ(C)= l o g 2 C γ + b γ o d d
式(1)中:C为通道的数量;bγ为两个超参数,默认值为2;odd表示k为奇数,若为偶数则加1。
因此为了进一步提升YOLOv8的特征提取能力,并保证引入少量的参数量,在Bottleneck模块的基础上设计了一种特征选择融合模块MLC-Bottleneck,替换掉C2f中的Bottleneck模块,从而形成MLC-C2f特征选择融合模块,其结构如图6所示。
通过这样以添加少量参数量为代价,使YOLOv8在提取特征时更加关注重要的信息,提高了对小目标的感知能力,从而提高了检测小目标缺陷的准确率。
模型的轻量化是深度学习领域的一个重要发展方向,其目的是为在资源有限的环境下部署和运行模型,即在移动设备、边缘设备和物联网设备等性能有限的场景下发挥作用,提高模型的部署效率和性能。因此为提高YOLOv8的部署效率,在保证精度的情况下尽可能减小其参数的数量。
在原YOLOv8中,检测头的参数量较多,并且由于检测的图片较大,不利于其在资源有限的设备中部署。而参数量如此庞大的原因为YOLOv8所使用的是解耦头,其结构如图7所示,这种检测头能够使每个任务的学习目标得到充分的考虑,不会受到其他任务的干扰,即可以灵活地处理多个不同类型的任务,但是增加了模型的参数量和计算复杂度。
图7可以看出,在最后的检测中使用了6次的卷积运算,并且会遍历3次,而卷积层的计算量可表示为
FLOPs=2HW(CinK2+1)Cout
式(2)中:FLOPs为浮点运算次数,可用来表示算法模型的复杂程度;HW分别为卷积核的高度和宽度,CinCout分别为输入和输出的通道数;K则为卷积核的大小。
根据式(2)可以发现检测头中的卷积操作是造成检测头庞大的原因之一,但是在黑点缺陷的检测中,任务类别仅为1,解耦头发挥不出设计的性能,有些参数是无效的[15]。为此设计一种共享参数检测头(shared parameter detect, SP-Detect),通过使用共享参数的方法,降低检测头的参数量,其结构如图8所示,通过共享参数层将所预测的结构进行输出,而不影响黑点缺陷的检测精度,这样一来降低所用的参数量,进而使YOLOv8进一步轻量化。
经过上述所有的改进,基于YOLOv8的陶瓷杯小目标缺陷检测算法的结构如图9所示。
在现有的数据集中并未发现有关陶瓷杯缺陷的数据集,因此本文中也对数据集进行新建。根据相关要求陶瓷杯缺陷的检测精度为0.1 mm,为保证其清晰采集,通过下式进行计算工业相机的分辨率,公式为
分辨率= ×3
经过计算选用500万像素的工业相机,并对陶瓷杯缺陷进行采集,通过采集得到了508张图像,共966个黑点缺陷实例,随后通过Labelimg软件对图像进行标注,并按照8∶2的比例将数据集分为了训练集和测试集。
硬件部分:CPU为Intel(R) Core(TM) i5-9500H CPU @3.0 GHz,内存为16 GB,显卡为NVIDIA GeForce GTX 2060Super,显存为8 GB。
软件部分:操作系统为Win10,部署环境为Python3.9.13,训练框架为PyTorch1.10.2,加速计算框架为CUDA10.2。
在本次实验中,评价指标使用mAP(mean average precision)以及参数量(parameters)两个主要的指标,其中mAP值用来衡量神经网络的检测精度。在计算以上指标时又使用了准确率P(precision)、召回率R(recall),计算PR所需参数的含义如表2混淆矩阵所示。
计算公式为
P= T P T P + F P
R= T P T P + F N
根据计算得到的PR,生成P-R图像,AP为图像与坐标轴所围成的面积,计算公式为
AP= 0 1 PdR
mAP也由此图像进行计算,由AP取平均值而来,计算公式为
mAP= i = 0 N A P i N
为了保证实验的准确性,本次实验在相同的环境及设置下,各个项目进行多次实验并求取平均值,训练次数均为300次,具体数据如表3所示。
通过表3可以看出,所改进模块在检测效果上呈上升的趋势,本算法的mAP达到了97.5%,相较于YOLOv8算法,其mAP提升了20.9%,参数量下降了66%。在具体的检测效果中可以看出检测后的缺陷几乎都可以识别与定位,且检测缺陷的尺度范围更广。而个别未检测到的缺陷是由于图像的采集不清晰导致,该问题可在实际生产中添加光源来避免,以进一步提升检测的全面性。因此,本文中所提出的算法,在检测精度以及参数量上都有着明显的优势,参数量优势代表其可以嵌入移动设备或终端设备中进行使用,能够有效节约算力,降低生产的成本投入。同时该算法具有良好的可扩展性,能够适应不同规模和复杂度的目标检测任务,这为小目标检测提供了一种通用的解决方案。
为验证所提出的算法在实际中的表现,开发了如图10所示的自动检测实验平台,其主要由传送带、工业相机、UR5机器人、光电开关组成。在实验平台中,传送带负责将陶瓷杯进行移动,工业相机对陶瓷杯面进行图像采集,UR5机器人则负责将带有缺陷的陶瓷杯进行剔除。
为满足工业缺陷检测实时性要求,选用了具有高数据传输速率和稳定性、支持互联网远程访问控制的 Modbus TCP 通讯协议,并使用 Python 语言结合该协议开发了自动控制系统。系统运行时,传送带移动陶瓷杯,光电开关检测到杯子到达图像采集区发出信号,此时开关量采集模块转换信号传至 PC 端,PC 端控制工业相机采集图像,并将图像送至改进算法中检测,若有缺陷则输出“NG”信号至 UR5 机器人剔除,无缺陷则进入下一步工序。
为进一步证明本文算法及系统的优越性,对检测产品的速度与人工检测进行了对比。经调查,一名熟练的检测人员检测一个产品的平均时间约为3.2 s(主要检测陶瓷杯圆周外表面),而在实验平台中将通过三台工业相机对陶瓷杯进行360°的图像采集,算法对生成的三张高分辨率图像进行检测。通过使用50只陶瓷杯进行检测实验,本文算法检测每一只杯子(三张高分辨率图像)的时间如图11所示。
图11中可看出算法检测每只杯子的时间是有波动的,但总体相差不大,而造成这种现象的原因是由于每只杯子上缺陷的多少是不一致的,取其平均值194.182 ms则近似为检测一只杯子的时间。另外采集图像的时间以及系统延迟时间之和约为1.5 s,即该实验平台检测一个产品的时间约为1.7 s,检测时间相较于人工降低了1.5 s,单只陶瓷杯的检测效率提升了46.9%,并可将缺陷的详细信息进行标记存储在系统中。此外,实验平台可在条件允许的情况下一直进行工作,其检测效率是远远高于人工的。
通过该实验的验证,表明所提出的算法和实验平台能够满足检测的要求,并能精确的将带有缺陷的陶瓷杯进行分离,这为后续设计陶瓷杯缺陷检测设备打下基础。
(1)提出了基于深度学习的陶瓷小目标缺陷检测算法,在其中研究了一种切片训练的方法,加快了训练时的速度,降低了近一倍的显卡内存占用率。另外添加了小目标检测头,并设计了一种MLC-C2f特征选择融合模块,从而提高对小目标缺陷特征的感知能力。最后结合研制的共享参数检测头,算法中可学习参数的数量降低了66%,并以陶瓷杯为例,对小目标缺陷检测的精度提升了20.9%。
(2)研制了一套自动检测实验平台,通过融合三台工业相机信息,对产品进行360°检测,并使用Python语言开发了自动控制系统软件。结合本文算法,在陶瓷杯的检测中,对比人工检测,检测完成一只杯子所需要的时间由3.2 s降低至1.7 s,其检测效率提升了46.9%,极大地提升了检测效率和检测准确性。
  • 国家自然科学基金(52075306)
参考文献 引证文献
排序方式:
[1]
潘晓英, 贾凝心, 穆元震, 等. 小目标检测研究综述[J]. 中国图象图形学报, 2023, 28(9): 2587-2615.
Pan Xiaoying, Jia Ningxin, Mu Yuanzhen, et al. Survey of small object detection[J]. Journal of Image and Graphics, 2023, 28(9): 2587-2615.
[2]
员娇娇, 胡永利, 孙艳丰, 等. 基于深度学习的小目标检测方法综述[J]. 北京工业大学学报, 2021, 47(3): 293-302.
Yuan Jiaojiao, Hu Yongli, Sun Yanfeng, et al. Survey of small object detection methods based on deep learning[J]. Journal of Beijing University of Technology, 2021, 47(3): 293-302.
[3]
庄建军, 石潇愉. 一种基于深度学习的商标检测方法[J]. 科学技术与工程, 2023, 23(36): 15538-15544.
Zhuang Jianjun, Shi Xiaoyu. A trademark detection method based on deep learning[J]. Science Technology and Engineering, 2023, 23(36): 15538-15544.
[4]
Cao T, Song K, Xu L, et al. Balanced multi-scale target score network for ceramic tile surface defect detection[J]. Measurement, 2024, 224: 113914.
[5]
邓钢, 赵庆华, 祁光威, 等. 基于YOLOv5的铝合金型材表面缺陷检测方法研究[J]. 现代制造工程, 2023(11): 120-128.
Deng Gang, Zhao Qinghua, Qi Guangwei, et al. Research on aluminum profile surface defects detection method based on YOLOv5[J]. Modern Manufacturing Engineering, 2023(11): 120-128.
[6]
王宏, 韩晨, 袁伯阳, 等. 基于改进YOLOv5的拥挤行人检测算法[J]. 科学技术与工程, 2023, 23(27): 11730-11738.
Wang Hong, Han Chen, Yuan Boyang, et al. Crowded pedestrian detection algorithm based on improved YOLOv5[J]. Modern Manufacturing Engineering, 2023, 23(27): 11730-11738.
[7]
洪腾蛟, 丁凤娟, 王鹏, 等. 深度学习在轴承故障诊断领域的应用研究[J]. 科学技术与工程, 2021, 21(22): 9203-9211.
Hong Tengjiao, Ding Fengjuan, Wang Peng, et al. Application of deep learning in bearing fault diagnosis[J]. Modern Manufacturing Engineering, 2021, 21(22): 9203-9211.
[8]
金映谷, 张涛, 杨亚宁, 等. 基于深度学习的产品缺陷检测方法综述[J]. 大连民族大学学报, 2020, 22(5): 420-427.
Jin Yinggu, Zhang Tao, Yang Yaning, et al. Review of product defect detection methods based on deep learning[J]. Journal of Dalian Minzu University, 2020, 22(5): 420-427.
[9]
Edmon J, Divvala S, Girshick R, et al. You only look once: unified, real-time object detection[C]// Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE, 2016: 779-788.
[10]
Liu W, Anguelov D, Erhan D, et al. SSD: single shot multibox detector[C]// Computer Vision-ECCV 2016. Cham: Springer, 2016: 21-37.
[11]
Ren S, He K, Girshick R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(6): 1137-1149.
[12]
王佳皓, 徐树公, 陆恒杰. 基于快速下采样的轻量化网络设计方法及人脸识别应用[J]. 电子学报, 2023, 51(8): 2226-2237.
Wang Jiahao, Xu Shugong, Lu Hengjie. Lightweight network design and application for face recognition based on fast down-sampling[J]. Acta Electronica Sinica, 2023, 51(8): 2226-2237.
[13]
何永春, 申永伟, 吴涛, 等. 基于注意力机制的多尺度仪表检测[J]. 科学技术与工程, 2021, 21(31): 13430-13438.
He Yongchun, Shen Yongwei, Wu Tao, et al. Multi-scale instrument detection based on attention mechanism[J]. Science Technology and Engineering, 2021, 21(31): 13430-13438.
[14]
Wan D, Lu R, Shen S, et al. Mixed local channel attention for object detection[J]. Engineering Applications of Artificial Intelligence, 2023, 123: 106442.
[15]
周颖, 颜毓泽, 陈海永, 等. 基于改进YOLOv8的光伏电池缺陷检测[J]. 激光与光电子学进展, 2024, 61(8): 245-255.
Zhou Ying, Yan Shuze, Chen Haiyong, et al. Defect detection of photovoltaic cells based on improved YOLOV8[J]. Laser & Optoelectronics Progress, 2024, 61(8): 245-255.
2025年第25卷第11期
PDF下载
218
75
引用本文
BibTeX
文章信息
doi: 10.12404/j.issn.1671-1815.2403574
  • 接收时间:2024-05-15
  • 首发时间:2025-07-09
  • 出版时间:2025-04-18
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2024-05-15
  • 修回日期:2024-08-10
基金
国家自然科学基金(52075306)
作者信息
    山东理工大学机械工程学院, 淄博 255049

通讯作者:

* 杨先海(1963—),男,汉族,山东邹平人,教授,博士研究生导师。研究方向:机器视觉、机械创新设计。E-mail:
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/kxjsygc/CN/10.12404/j.issn.1671-1815.2403574
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏