Article(id=1263514357103715295, tenantId=1146029695717560320, journalId=1263187241531621409, issueId=1263514351571428296, articleNumber=null, orderNo=null, doi=10.11996/JG.j.2095-302X.2026010017, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=research-article, receivedDate=1740672000000, receivedDateStr=2025-02-28, revisedDate=null, revisedDateStr=null, acceptedDate=1750608000000, acceptedDateStr=2025-06-23, onlineDate=1779174897695, onlineDateStr=2026-05-19, pubDate=1772208000000, pubDateStr=2026-02-28, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1779174897695, onlineIssueDateStr=2026-05-19, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1779174897695, creator=13701087609, updateTime=1779174897695, updator=13701087609, issue=Issue{id=1263514351571428296, tenantId=1146029695717560320, journalId=1263187241531621409, year='2026', volume='47', issue='1', pageStart='1', pageEnd='233', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=1, specialIssue=null, createTime=1779174896376, creator=13701087609, updateTime=1779174963943, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1263514635077039012, tenantId=1146029695717560320, journalId=1263187241531621409, issueId=1263514351571428296, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1263514635077039013, tenantId=1146029695717560320, journalId=1263187241531621409, issueId=1263514351571428296, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=17, endPage=28, ext={EN=ArticleExt(id=1263514359247004649, articleId=1263514357103715295, tenantId=1146029695717560320, journalId=1263187241531621409, language=EN, title=A vehicle damage classification model incorporating dual attention and weighted dynamic convolution, columnId=1263514354654262248, journalTitle=Journal of Graphics, columnName=Image Processing and Computer Vision, runingTitle=null, highlight=null, articleAbstract=

To address the challenges of morphological similarity and the resulting difficulty in classifying vehicle damage images uploaded by clients for auto insurance claims, a model named ResAWDNet was proposed for vehicle damage classification. Firstly, to effectively augment the model’s capacity for extracting damage features, the traditional down sampling operation was replaced with weighted dynamic convolution. This approach dynamically adjusted the weights of convolutional kernels based on the input features, thereby enhancing the model’s adaptability to features of varying scales and orientations. As a result, it enabled more precise capture of the subtle differences in vehicle damage. Secondly, to ensure that the model could concentrate on the salient discriminative regions and feature channels within the images, a dual attention mechanism was embedded after the convolutional layers of the backbone network. This mechanism concurrently learned the important weights in both spatial and channel dimensions, significantly enhancing the model’s ability to capture crucial information. Consequently, it further enhanced the decision-making accuracy of the model in the task of vehicle damage classification. Finally, experimental validation was conducted based on a dataset of vehicle damage images sourced from real accident cases. The experimental results demonstrated that the ResAWDNet model was feasible and offered significant advantages for vehicle damage classification tasks, achieving an accuracy rate of 73.79%. Compared with baseline models, ResAWDNet achieved higher accuracy in classifying multiple types of damages, robustly validating the effectiveness of the proposed model.

, correspAuthors=null, authorNote=null, correspAuthorsNote=
WANG Qianming,E-mail:
, 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=Yongjie ZHAI, Zixuan WANG, Zhenqi ZHANG, Xunqi ZHOU, Qianming WANG), CN=ArticleExt(id=1263514364343083054, articleId=1263514357103715295, tenantId=1146029695717560320, journalId=1263187241531621409, language=CN, title=融合双重注意力与加权动态卷积的车辆损伤分类模型, columnId=1263514354893337578, journalTitle=图学学报, columnName=图像处理与计算机视觉, runingTitle=null, highlight=null, articleAbstract=

针对车险理赔客户上传的车辆损伤图像中存在损伤类型形态相似、分类困难的问题,提出了一种适用于车辆损伤分类的模型ResAWDNet。首先,为有效增强模型对损伤特征的提取能力,使用加权动态卷积代替原有的下采样操作,依据输入特征动态调整卷积核权重,提高模型对不同尺度和方向特征的适应性,从而更准确地捕捉损伤的细微差异。其次,为了使模型关注图像中的显著性判别区域和特征通道,在主干网络的卷积层后嵌入了双重注意力机制,同时学习空间和通道维度上的重要权重,提升模型对关键信息的捕捉能力,进一步提升模型在损伤分类任务中的决策准确性。最后,基于真实事故案例的车辆损伤图片数据集进行实验验证。实验结果表明,ResAWDNet模型在车辆损伤分类任务中切实可行且优势显著,整体分类准确率达到73.79%。与基线模型相比,ResAWDNet在多类损伤类型的分类上均展现出更高的准确率,有力地证明了该模型的有效性。

, correspAuthors=null, authorNote=null, correspAuthorsNote=
王乾铭,E-mail:
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=D3O+RzQ9kRXAzLtkEogbpA==, magXml=tj9PY3SEU5EamKRGk4qwSg==, pdfUrl=null, pdf=XArGtMagVXYAYfeoT6MTKg==, pdfFileSize=16395125, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=BFn+/l2Vs5lB+G0P9emWKA==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=nAVj6kTPxk6OuOSceBBp8g==, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=翟永杰, 王紫萱, 张祯琪, 周迅琪, 王乾铭)}, authors=[Author(id=1263550825205477485, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, orderNo=0, 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=1263550826346328183, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550825205477485, language=EN, stringName=Yongjie ZHAI, firstName=Yongjie, middleName=null, lastName=ZHAI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1263550826899976316, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550825205477485, language=CN, stringName=翟永杰, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=华北电力大学自动化系, 河北 保定 071003, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1263550824823795811, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, xref=null, ext=[AuthorCompanyExt(id=1263550824857350244, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China), AuthorCompanyExt(id=1263550824861544549, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华北电力大学自动化系, 河北 保定 071003)])]), Author(id=1263550829362032776, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, 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=1263550829945041037, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550829362032776, language=EN, stringName=Zixuan WANG, firstName=Zixuan, middleName=null, lastName=WANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1263550830637101202, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550829362032776, language=CN, stringName=王紫萱, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=华北电力大学自动化系, 河北 保定 071003, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1263550824823795811, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, xref=null, ext=[AuthorCompanyExt(id=1263550824857350244, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China), AuthorCompanyExt(id=1263550824861544549, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华北电力大学自动化系, 河北 保定 071003)])]), Author(id=1263550831375298716, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, 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=1263550831970889887, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550831375298716, language=EN, stringName=Zhenqi ZHANG, firstName=Zhenqi, middleName=null, lastName=ZHANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1263550835229864114, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550831375298716, language=CN, stringName=张祯琪, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=华北电力大学自动化系, 河北 保定 071003, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1263550824823795811, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, xref=null, ext=[AuthorCompanyExt(id=1263550824857350244, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China), AuthorCompanyExt(id=1263550824861544549, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华北电力大学自动化系, 河北 保定 071003)])]), Author(id=1263550837301850305, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, orderNo=3, 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=1263550840191725768, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550837301850305, language=EN, stringName=Xunqi ZHOU, firstName=Xunqi, middleName=null, lastName=ZHOU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1263550841173192911, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550837301850305, language=CN, stringName=周迅琪, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=华北电力大学自动化系, 河北 保定 071003, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1263550824823795811, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, xref=null, ext=[AuthorCompanyExt(id=1263550824857350244, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China), AuthorCompanyExt(id=1263550824861544549, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华北电力大学自动化系, 河北 保定 071003)])]), Author(id=1263550843169681624, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=qianmingwang@ncepu.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1263550844046291165, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550843169681624, language=EN, stringName=Qianming WANG, firstName=Qianming, middleName=null, lastName=WANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1263550845325553894, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, authorId=1263550843169681624, language=CN, stringName=王乾铭, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=华北电力大学自动化系, 河北 保定 071003, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1263550824823795811, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, xref=null, ext=[AuthorCompanyExt(id=1263550824857350244, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China), AuthorCompanyExt(id=1263550824861544549, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华北电力大学自动化系, 河北 保定 071003)])])], keywords=[Keyword(id=1263550848999764207, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, orderNo=1, keyword=intelligent damage assessment), Keyword(id=1263550849595355379, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, orderNo=2, keyword=image classification), Keyword(id=1263550851923194105, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, orderNo=3, keyword=deep learning), Keyword(id=1263550853252788479, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, orderNo=4, keyword=attention mechanism), Keyword(id=1263550853860962564, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, orderNo=5, keyword=dynamic convolution), Keyword(id=1263550855752593674, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, orderNo=1, keyword=智能定损), Keyword(id=1263550856134275345, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, orderNo=2, keyword=图像分类), Keyword(id=1263550856734060824, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, orderNo=3, keyword=深度学习), Keyword(id=1263550857115742492, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, orderNo=4, keyword=注意力机制), Keyword(id=1263550857908465957, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, orderNo=5, keyword=动态卷积)], refs=[Reference(id=1263550879332971024, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2019, volume=33, issue=3, pageStart=73, pageEnd=77, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=赵子豪, 申颖, 李薇, journalName=保险职业学院学报, refType=null, unstructuredReference=赵子豪, 申颖, 李薇. 基于图像识别的车辆智能定损应用研究[J]. 保险职业学院学报, 2019, 33(3): 73-77., articleTitle=基于图像识别的车辆智能定损应用研究, refAbstract=null), Reference(id=1263550879588823575, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2019, volume=33, issue=3, pageStart=73, pageEnd=77, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=ZHAO Z H, SHEN Y, LI W, journalName=Journal of Insurance Professional College, refType=null, unstructuredReference=ZHAO Z H, SHEN Y, LI W. Application and value research about apps of vehicle survey and loss assessment based on image recognition[J]. Journal of Insurance Professional College, 2019, 33(3): 73-77 (in Chinese)., articleTitle=Application and value research about apps of vehicle survey and loss assessment based on image recognition, refAbstract=null), Reference(id=1263550879924367899, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=45, issue=5, pageStart=930, pageEnd=940, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=翟永杰, 李佳蔚, 陈年昊, journalName=图学学报, refType=null, unstructuredReference=翟永杰, 李佳蔚, 陈年昊, . 融合改进Transformer的车辆部件检测方法[J]. 图学学报, 2024, 45(5): 930-940., articleTitle=融合改进Transformer的车辆部件检测方法, refAbstract=null), Reference(id=1263550880176026141, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=45, issue=5, pageStart=930, pageEnd=940, url=null, language=null, rfNumber=[2], rfOrder=3, authorNames=ZHAI Y J, LI J W, CHEN N H, journalName=Journal of Graphics, refType=null, unstructuredReference=ZHAI Y J, LI J W, CHEN N H, et al. The vehicle parts detection method enhanced with Transformer integration[J]. Journal of Graphics, 2024, 45(5): 930-940 (in Chinese)., articleTitle=The vehicle parts detection method enhanced with Transformer integration, refAbstract=null), Reference(id=1263550880423490082, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=45, issue=4, pageStart=770, pageEnd=778, url=null, language=null, rfNumber=[3], rfOrder=4, authorNames=武兵, 田莹, journalName=图学学报, refType=null, unstructuredReference=武兵, 田莹. 基于注意力机制的多尺度道路损伤检测算法研究[J]. 图学学报, 2024, 45(4): 770-778., articleTitle=基于注意力机制的多尺度道路损伤检测算法研究, refAbstract=null), Reference(id=1263550880587067940, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=45, issue=4, pageStart=770, pageEnd=778, url=null, language=null, rfNumber=[3], rfOrder=5, authorNames=WU B, TIAN Y, journalName=Journal of Graphics, refType=null, unstructuredReference=WU B, TIAN Y. Research on multi-scale road damage detection algorithm based on attention mechanism[J]. Journal of Graphics, 2024, 45(4): 770-778 (in Chinese)., articleTitle=Research on multi-scale road damage detection algorithm based on attention mechanism, refAbstract=null), Reference(id=1263550880796783147, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2022, volume=12, issue=1, pageStart=10689, pageEnd=null, url=null, language=null, rfNumber=[4], rfOrder=6, authorNames=LIU Q, HUANG X H, SHAO X Y, journalName=Scientific Reports, refType=null, unstructuredReference=LIU Q, HUANG X H, SHAO X Y, et al. Industrial cylinder liner defect detection using a transformer with a block division and mask mechanism[J]. Scientific Reports, 2022, 12(1): 10689., articleTitle=Industrial cylinder liner defect detection using a transformer with a block division and mask mechanism, refAbstract=null), Reference(id=1263550882524836399, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2016, volume=3, issue=3, pageStart=303, pageEnd=310, url=null, language=null, rfNumber=[5], rfOrder=7, authorNames=PARK J K, KWON B K, PARK J H, journalName=International Journal of Precision Engineering and Manufacturing-Green Technology, refType=null, unstructuredReference=PARK J K, KWON B K, PARK J H, et al. Machine learning-based imaging system for surface defect inspection[J]. International Journal of Precision Engineering and Manufacturing-Green Technology, 2016, 3(3): 303-310., articleTitle=Machine learning-based imaging system for surface defect inspection, refAbstract=null), Reference(id=1263550882709385777, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2021, volume=11, issue=23, pageStart=11459, pageEnd=null, url=null, language=null, rfNumber=[6], rfOrder=8, authorNames=WU S Q, ZHAO S Y, ZHANG Q Q, journalName=Applied Sciences, refType=null, unstructuredReference=WU S Q, ZHAO S Y, ZHANG Q Q, et al. Steel surface defect classification based on small sample learning[J]. Applied Sciences, 2021, 11(23): 11459., articleTitle=Steel surface defect classification based on small sample learning, refAbstract=null), Reference(id=1263550882843603508, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=24, issue=4, pageStart=041005, pageEnd=null, url=null, language=null, rfNumber=[7], rfOrder=9, authorNames=LIU W, QIU J, WANG Y, journalName=Journal of Computing and Information Science in Engineering, refType=null, unstructuredReference=LIU W, QIU J, WANG Y, et al. Multiscale feature fusion convolutional neural network for surface damage detection in retired steel shafts[J]. Journal of Computing and Information Science in Engineering, 2024, 24(4): 041005.., articleTitle=Multiscale feature fusion convolutional neural network for surface damage detection in retired steel shafts, refAbstract=null), Reference(id=1263550883325948477, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[8], rfOrder=10, authorNames=王瑞芳, journalName=基于字典学习的图像分类算法研究, refType=null, unstructuredReference=王瑞芳. 基于字典学习的图像分类算法研究[D]. 重庆: 重庆邮电大学, 2020., articleTitle=null, refAbstract=null), Reference(id=1263550883745378881, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[8], rfOrder=11, authorNames=WANG R F, journalName=Research on image classification algorithm based on dictionary learning, refType=null, unstructuredReference=WANG R F. Research on image classification algorithm based on dictionary learning[D]. Chongqing: Chongqing University of Posts and Telecommunications, 2020 (in Chinese)., articleTitle=null, refAbstract=null), Reference(id=1263550884101894724, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2021, volume=42, issue=4, pageStart=572, pageEnd=580, url=null, language=null, rfNumber=[9], rfOrder=12, authorNames=张鹏飞, 石志良, 李晓垚, journalName=图学学报, refType=null, unstructuredReference=张鹏飞, 石志良, 李晓垚, . 基于深度学习的主轴承盖分类识别算法[J]. 图学学报, 2021, 42(4): 572-580., articleTitle=基于深度学习的主轴承盖分类识别算法, refAbstract=null), Reference(id=1263550884198363717, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2021, volume=42, issue=4, pageStart=572, pageEnd=580, url=null, language=null, rfNumber=[9], rfOrder=13, authorNames=ZHANG P F, SHI Z L, LI X Y, journalName=Journal of Graphics, refType=null, unstructuredReference=ZHANG P F, SHI Z L, LI X Y, et al. Classification algorithm of main bearing cap based on deep learning[J]. Journal of Graphics, 2021, 42(4): 572-580 (in Chinese)., articleTitle=Classification algorithm of main bearing cap based on deep learning, refAbstract=null), Reference(id=1263550884315804231, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[10], rfOrder=14, authorNames=董潇, journalName=卷积神经网络的图像分类优化算法研究, refType=null, unstructuredReference=董潇. 卷积神经网络的图像分类优化算法研究[D]. 淮南: 安徽理工大学, 2020., articleTitle=null, refAbstract=null), Reference(id=1263550884445827657, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[10], rfOrder=15, authorNames=DONG X, journalName=Research on image classification optimization algorithm of convolutional neural network, refType=null, unstructuredReference=DONG X. Research on image classification optimization algorithm of convolutional neural network[D]. Huainan: Anhui University of Science & Technology, 2020 (in Chinese)., articleTitle=null, refAbstract=null), Reference(id=1263550884638765643, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2021, volume=26, issue=5, pageStart=1030, pageEnd=1040, url=null, language=null, rfNumber=[11], rfOrder=16, authorNames=贺敏雪, 余烨, 程茹秋, journalName=中国图象图形学报, refType=null, unstructuredReference=贺敏雪, 余烨, 程茹秋. 特征增强策略驱动的车标识别[J]. 中国图象图形学报, 2021, 26(5): 1030-1040., articleTitle=特征增强策略驱动的车标识别, refAbstract=null), Reference(id=1263550884865258061, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2021, volume=26, issue=5, pageStart=1030, pageEnd=1040, url=null, language=null, rfNumber=[11], rfOrder=17, authorNames=HE M X, YU Y, CHENG R Q, journalName=Journal of Image and Graphics, refType=null, unstructuredReference=HE M X, YU Y, CHENG R Q. Vehicle logo recognition method based on feature enhancement[J]. Journal of Image and Graphics, 2021, 26(5): 1030-1040 (in Chinese)., articleTitle=Vehicle logo recognition method based on feature enhancement, refAbstract=null), Reference(id=1263550885045613135, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=6, issue=2, pageStart=633, pageEnd=646, url=null, language=null, rfNumber=[12], rfOrder=18, authorNames=ANANDA B, PUTRI R A, journalName=Journal of Computer Networks, Architecture and High Performance Computing, refType=null, unstructuredReference=ANANDA B, PUTRI R A. K-nearest neighbor algorithm and case base reasoning on xenia car damage detection expert system[J]. Journal of Computer Networks, Architecture and High Performance Computing, 2024, 6(2): 633-646., articleTitle=K-nearest neighbor algorithm and case base reasoning on xenia car damage detection expert system, refAbstract=null), Reference(id=1263550885213385297, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=1, pageEnd=6, url=null, language=null, rfNumber=[13], rfOrder=19, authorNames=MISHRA S, KAMAL D, SENTHIL KUMAR K, journalName=2024 IEEE International Students' Conference on Electrical, Electronics and Computer Science, refType=null, unstructuredReference=MISHRA S, KAMAL D, SENTHIL KUMAR K. Vehicle damage identification using deep learning techniques[C]// 2024 IEEE International Students' Conference on Electrical, Electronics and Computer Science. New York: IEEE Press, 2024: 1-6., articleTitle=Vehicle damage identification using deep learning techniques, refAbstract=null), Reference(id=1263550885360185938, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2025, volume=26, issue=4, pageStart=4717, pageEnd=4730, url=null, language=null, rfNumber=[14], rfOrder=20, authorNames=PENG J B, DONG S B, YUAN H, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=PENG J B, DONG S B, YUAN H, et al. Car damage detection based on multi-view fusion and alignment: dataset and method[J]. IEEE Transactions on Intelligent Transportation Systems, 2025, 26(4): 4717-4730., articleTitle=Car damage detection based on multi-view fusion and alignment: dataset and method, refAbstract=null), Reference(id=1263550885460849236, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=101, pageEnd=107, url=null, language=null, rfNumber=[15], rfOrder=21, authorNames=SHUBHAM, BANERJEE D, journalName=The 4th International Conference on Technological Advancements in Computational Sciences, refType=null, unstructuredReference=SHUBHAM, BANERJEE D. Robust car damage identification through CNN and SVM techniques[C]// The 4th International Conference on Technological Advancements in Computational Sciences. New York: IEEE Press, 2024: 101-107., articleTitle=Robust car damage identification through CNN and SVM techniques, refAbstract=null), Reference(id=1263550887201485398, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[16], rfOrder=22, authorNames=王心旷, journalName=基于深度学习的车辆外观损伤识别及其图像生成方法研究, refType=null, unstructuredReference=王心旷. 基于深度学习的车辆外观损伤识别及其图像生成方法研究[D]. 合肥: 中国科学技术大学, 2024., articleTitle=null, refAbstract=null), Reference(id=1263550887482503768, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[16], rfOrder=23, authorNames=WANG X K, journalName=Research on car exterior damage recognition and image generation based on deep learning, refType=null, unstructuredReference=WANG X K. Research on car exterior damage recognition and image generation based on deep learning[D]. Hefei: University of Science and Technology of China, 2024 (in Chinese)., articleTitle=null, refAbstract=null), Reference(id=1263550887591555674, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=770, pageEnd=778, url=null, language=null, rfNumber=[17], rfOrder=24, authorNames=HE K M, ZHANG X Y, REN S Q, journalName=2016 IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016: 770-778., articleTitle=Deep residual learning for image recognition, refAbstract=null), Reference(id=1263550887696413276, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2017, volume=39, issue=6, pageStart=1137, pageEnd=1149, url=null, language=null, rfNumber=[18], rfOrder=25, authorNames=REN S Q, HE K M, GIRSHICK R, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149., articleTitle=Faster R-CNN: towards real-time object detection with region proposal networks, refAbstract=null), Reference(id=1263550887792882270, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/abs/1412.7062., language=null, rfNumber=[19], rfOrder=26, authorNames=CHEN L C, PAPANDREOU G, KOKKINOS I, journalName=null, refType=null, unstructuredReference=CHEN L C, PAPANDREOU G, KOKKINOS I, et al. Semantic image segmentation with deep convolutional nets and fully connected CRFs[EB/OL]. [2024-12-28]. https://arxiv.org/abs/1412.7062., articleTitle=Semantic image segmentation with deep convolutional nets and fully connected CRFs, refAbstract=null), Reference(id=1263550888044540512, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://openreview.net/forum?id=Hk99zCeAb, language=null, rfNumber=[20], rfOrder=27, authorNames=KARRAS T, AILA T, LAINE S, journalName=null, refType=null, unstructuredReference=KARRAS T, AILA T, LAINE S, et al. Progressive growing of GANs for improved quality, stability, and variation[EB/OL]. [2024-12-28]. https://openreview.net/forum?id=Hk99zCeAb., articleTitle=Progressive growing of GANs for improved quality, stability, and variation, refAbstract=null), Reference(id=1263550888333947490, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=18, issue=10, pageStart=2727, pageEnd=2737, url=null, language=null, rfNumber=[21], rfOrder=28, authorNames=顾正华, 刘嘎琼, 邵长斌, journalName=计算机科学与探索, refType=null, unstructuredReference=顾正华, 刘嘎琼, 邵长斌, . 深度检测方法中融合大小感受野机制的下采样算法[J]. 计算机科学与探索, 2024, 18(10): 2727-2737., articleTitle=深度检测方法中融合大小感受野机制的下采样算法, refAbstract=null), Reference(id=1263550888707240548, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=18, issue=10, pageStart=2727, pageEnd=2737, url=null, language=null, rfNumber=[21], rfOrder=29, authorNames=GU Z H, LIU G Q, SHAO C B, journalName=Journal of Frontiers of Technology, refType=null, unstructuredReference=GU Z H, LIU G Q, SHAO C B, et al. Downsampling algorithm with fusion of different receptive field sizes in deep detection methods[J]. Journal of Frontiers of Technology, 2024, 18(10): 2727-2737 (in Chinese)., articleTitle=Downsampling algorithm with fusion of different receptive field sizes in deep detection methods, refAbstract=null), Reference(id=1263550888816292453, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[22], rfOrder=30, authorNames=谢东升, journalName=基于深度学习的车辆智能定损算法研究, refType=null, unstructuredReference=谢东升. 基于深度学习的车辆智能定损算法研究[D]. 天津: 天津大学, 2019., articleTitle=null, refAbstract=null), Reference(id=1263550889214751334, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[22], rfOrder=31, authorNames=XIE D S, journalName=Research on vehicle intelligent damage location algorithm based on deep learning, refType=null, unstructuredReference=XIE D S. Research on vehicle intelligent damage location algorithm based on deep learning[D]. Tianjin: Tianjin University, 2019 (in Chinese)., articleTitle=null, refAbstract=null), Reference(id=1263550889420272232, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=7132, pageEnd=7141, url=null, language=null, rfNumber=[23], rfOrder=32, authorNames=HU J, SHEN L, SUN G, journalName=2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2018: 7132-7141., articleTitle=Squeeze-and-excitation networks, refAbstract=null), Reference(id=1263550889516741226, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=2017, pageEnd=2025, url=null, language=null, rfNumber=[24], rfOrder=33, authorNames=JADERBERG M, SIMONYAN K, ZISSERMAN A, journalName=The 29th International Conference on Neural Information Processing Systems, refType=null, unstructuredReference=JADERBERG M, SIMONYAN K, ZISSERMAN A. Spatial transformer networks[C]// The 29th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2015: 2017-2025., articleTitle=Spatial transformer networks, refAbstract=null), Reference(id=1263550889789370988, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=3, pageEnd=19, url=null, language=null, rfNumber=[25], rfOrder=34, authorNames=WOO S, PARK J, LEE J Y, journalName=The 15th European Conference on Computer Vision, refType=null, unstructuredReference=WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]// The 15th European Conference on Computer Vision. Cham: Springer, 2018: 3-19., articleTitle=CBAM: convolutional block attention module, refAbstract=null), Reference(id=1263550891584533102, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/abs/2410.13276, language=null, rfNumber=[26], rfOrder=35, authorNames=GAO Y, ZENG Z, DU D, journalName=null, refType=null, unstructuredReference=GAO Y, ZENG Z, DU D, et al. SeerAttention: learning intrinsic sparse attention in your LLM[EB/OL]. [2024-12-28]. https://arxiv.org/abs/2410.13276., articleTitle=SeerAttention: learning intrinsic sparse attention in your LLM, refAbstract=null), Reference(id=1263550891844579952, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=1, pageEnd=5, url=null, language=null, rfNumber=[27], rfOrder=36, authorNames=OUYANG D L, HE S, ZHANG G Z, journalName=ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing, refType=null, unstructuredReference=OUYANG D L, HE S, ZHANG G Z, et al. Efficient multi-scale attention module with cross-spatial learning[C]// ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing. New York: IEEE Press, 2023: 1-5., articleTitle=Efficient multi-scale attention module with cross-spatial learning, refAbstract=null), Reference(id=1263550891957826162, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=541, pageEnd=557, url=null, language=null, rfNumber=[28], rfOrder=37, authorNames=ZHANG H, ZU K K, LU J, journalName=The 16th Asian Conference on Computer Vision, refType=null, unstructuredReference=ZHANG H, ZU K K, LU J, et al. EPSANet: an efficient pyramid squeeze attention block on convolutional neural network[C]// The 16th Asian Conference on Computer Vision. Cham: Springer, 2022: 541-557., articleTitle=EPSANet: an efficient pyramid squeeze attention block on convolutional neural network, refAbstract=null), Reference(id=1263550892079460980, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=11531, pageEnd=11539, url=null, language=null, rfNumber=[29], rfOrder=38, authorNames=WANG Q L, WU B G, ZHU P F, journalName=2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=WANG Q L, WU B G, ZHU P F, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2020: 11531-11539., articleTitle=ECA-Net: efficient channel attention for deep convolutional neural networks, refAbstract=null), Reference(id=1263550892184318582, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2023, volume=131, issue=8, pageStart=2198, pageEnd=2218, url=null, language=null, rfNumber=[30], rfOrder=39, authorNames=CHEN L W, FU Y, WEI K X, journalName=International Journal of Computer Vision, refType=null, unstructuredReference=CHEN L W, FU Y, WEI K X, et al. Instance segmentation in the dark[J]. International Journal of Computer Vision, 2023, 131(8): 2198-2218., articleTitle=Instance segmentation in the dark, refAbstract=null), Reference(id=1263550892293370488, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2017, volume=null, issue=null, pageStart=3156, pageEnd=3164, url=null, language=null, rfNumber=[31], rfOrder=40, authorNames=WANG F, JIANG M Q, QIAN C, journalName=2017 IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=WANG F, JIANG M Q, QIAN C, et al. Residual attention network for image classification[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2017: 3156-3164., articleTitle=Residual attention network for image classification, refAbstract=null), Reference(id=1263550892540834424, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2023, volume=24, issue=7, pageStart=7202, pageEnd=7214, url=null, language=null, rfNumber=[32], rfOrder=41, authorNames=WANG X K, LI W J, WU Z C, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=WANG X K, LI W J, WU Z C. CarDD: a new dataset for vision-based car damage detection[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(7): 7202-7214., articleTitle=CarDD: a new dataset for vision-based car damage detection, refAbstract=null), Reference(id=1263550892696023674, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=3183, pageEnd=3192, url=null, language=null, rfNumber=[33], rfOrder=42, authorNames=ZHANG Z Z, LAN C L, ZENG W J, journalName=2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=ZHANG Z Z, LAN C L, ZENG W J, et al. Relation-aware global attention for person re-identification[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2020: 3183-3192., articleTitle=Relation-aware global attention for person re-identification, refAbstract=null), Reference(id=1263550893006402172, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=178, issue=null, pageStart=108784, pageEnd=null, url=null, language=null, rfNumber=[34], rfOrder=43, authorNames=HUANG H J, CHEN Z G, ZOU Y, journalName=Computers in Biology and Medicine, refType=null, unstructuredReference=HUANG H J, CHEN Z G, ZOU Y, et al. Channel prior convolutional attention for medical image segmentation[J]. Computers in Biology and Medicine, 2024, 178: 108784., articleTitle=Channel prior convolutional attention for medical image segmentation, refAbstract=null), Reference(id=1263550893291614846, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2017, volume=null, issue=null, pageStart=618, pageEnd=626, url=null, language=null, rfNumber=[35], rfOrder=44, authorNames=SELVARAJU R R, COGSWELL M, DAS A, journalName=2017 IEEE International Conference on Computer Vision, refType=null, unstructuredReference=SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-cam: visual explanations from deep networks via gradient-based localization[C]// 2017 IEEE International Conference on Computer Vision. New York: IEEE Press, 2017: 618-626., articleTitle=Grad-cam: visual explanations from deep networks via gradient-based localization, refAbstract=null), Reference(id=1263550893404861056, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2012, volume=null, issue=null, pageStart=1097, pageEnd=1105, url=null, language=null, rfNumber=[36], rfOrder=45, authorNames=KRIZHEVSKY A, SUTSKEVER I, HINTON G E, journalName=The 26th International Conference on Neural Information Processing Systems, refType=null, unstructuredReference=KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]// The 26th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2012: 1097-1105., articleTitle=ImageNet classification with deep convolutional neural networks, refAbstract=null), Reference(id=1263550893631353473, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=1, pageEnd=9, url=null, language=null, rfNumber=[37], rfOrder=46, authorNames=SZEGEDY C, LIU W, JIA Y Q, journalName=2015 IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2015: 1-9., articleTitle=Going deeper with convolutions, refAbstract=null), Reference(id=1263550893954314883, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/abs/1704.04861, language=null, rfNumber=[38], rfOrder=47, authorNames=HOWARD A G, ZHU M L, CHEN B, journalName=null, refType=null, unstructuredReference=HOWARD A G, ZHU M L, CHEN B, et al. MobileNets: efficient convolutional neural networks for mobile vision applications[EB/OL]. [2024-12-28]. https://arxiv.org/abs/1704.04861., articleTitle=MobileNets: efficient convolutional neural networks for mobile vision applications, refAbstract=null), Reference(id=1263550894134669957, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=6848, pageEnd=6856, url=null, language=null, rfNumber=[39], rfOrder=48, authorNames=ZHANG X Y, ZHOU X Y, LIN M X, journalName=2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=ZHANG X Y, ZHOU X Y, LIN M X, et al. ShuffleNet: an extremely efficient convolutional neural network for mobile devices[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2018: 6848-6856., articleTitle=ShuffleNet: an extremely efficient convolutional neural network for mobile devices, refAbstract=null), Reference(id=1263550894239527559, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2017, volume=null, issue=null, pageStart=2261, pageEnd=2269, url=null, language=null, rfNumber=[40], rfOrder=49, authorNames=HUANG G, LIU Z, VAN DER MAATEN L, journalName=2017 IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=HUANG G, LIU Z, VAN DER MAATEN L, et al. Densely connected convolutional networks[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2017: 2261-2269., articleTitle=Densely connected convolutional networks, refAbstract=null), Reference(id=1263550894424076937, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://proceedings.mlr.press/v97/tan19a.html, language=null, rfNumber=[41], rfOrder=50, authorNames=TAN M X, LE Q, journalName=null, refType=null, unstructuredReference=TAN M X, LE Q. EfficientNet: rethinking model scaling for convolutional neural networks[EB/OL]. [2024-12-28]. https://proceedings.mlr.press/v97/tan19a.html., articleTitle=EfficientNet: rethinking model scaling for convolutional neural networks, refAbstract=null), Reference(id=1263550896286347915, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2023, volume=34, issue=11, pageStart=9562, pageEnd=9567, url=null, language=null, rfNumber=[42], rfOrder=51, authorNames=XU J, PAN Y, PAN X L, journalName=IEEE Transactions on Neural Networks and Learning Systems, refType=null, unstructuredReference=XU J, PAN Y, PAN X L, et al. RegNet: self-regulated network for image classification[J]. IEEE Transactions on Neural Networks and Learning Systems, 2023, 34(11): 9562-9567., articleTitle=RegNet: self-regulated network for image classification, refAbstract=null), Reference(id=1263550896487674509, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://proceedings.mlr.press/v139/tan21a, language=null, rfNumber=[43], rfOrder=52, authorNames=TAN M X, LE Q, journalName=null, refType=null, unstructuredReference=TAN M X, LE Q. EfficientNetV2:smaller models and faster training[EB/OL]. [2024-12-28]. https://proceedings.mlr.press/v139/tan21a., articleTitle=EfficientNetV2:smaller models and faster training, refAbstract=null), Reference(id=1263550896747721359, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=12021, pageEnd=12031, url=null, language=null, rfNumber=[44], rfOrder=53, authorNames=CHEN J R, KAO S H, HE H, journalName=2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=CHEN J R, KAO S H, HE H, et al. Run, don't walk: chasing higher FLOPS for faster neural networks[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2023: 12021-12031., articleTitle=Run, don't walk: chasing higher FLOPS for faster neural networks, refAbstract=null), Reference(id=1263550897049711249, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=11953, pageEnd=11965, url=null, language=null, rfNumber=[45], rfOrder=54, authorNames=DING X H, ZHANG X Y, HAN J G, journalName=2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=DING X H, ZHANG X Y, HAN J G, et al. Scaling up your kernels to 31×31: revisiting large kernel design in CNNs[C]// 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2022: 11953-11965., articleTitle=Scaling up your kernels to 31×31: revisiting large kernel design in CNNs, refAbstract=null), Reference(id=1263550897297175188, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=5694, pageEnd=5703, url=null, language=null, rfNumber=[46], rfOrder=55, authorNames=MA X, DAI X Y, BAI Y, journalName=2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=MA X, DAI X Y, BAI Y, et al. Rewrite the stars[C]// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2024: 5694-5703., articleTitle=Rewrite the stars, refAbstract=null), Reference(id=1263550897464947350, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://openreview.net/forum?id=YicbFdNTTy, language=null, rfNumber=[47], rfOrder=56, authorNames=DOSOVITSKIY A, BEYER L, KOLESNIKOV A, journalName=null, refType=null, unstructuredReference=DOSOVITSKIY A, BEYER L, KOLESNIKOV A, et al. An image is worth 16x16 words:transformers for image recognition at scale[EB/OL]. [2024-12-28]. https://openreview.net/forum?id=YicbFdNTTy., articleTitle=An image is worth 16x16 words:transformers for image recognition at scale, refAbstract=null), Reference(id=1263550897620136600, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=9992, pageEnd=10002, url=null, language=null, rfNumber=[48], rfOrder=57, authorNames=LIU Z, LIN Y T, CAO Y, journalName=2021 IEEE/CVF International Conference on Computer Vision, refType=null, unstructuredReference=LIU Z, LIN Y T, CAO Y, et al. Swin transformer: hierarchical vision transformer using shifted windows[C]// 2021 IEEE/CVF International Conference on Computer Vision. New York: IEEE Press, 2021: 9992-10002., articleTitle=Swin transformer: hierarchical vision transformer using shifted windows, refAbstract=null), Reference(id=1263550897724994202, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://openreview.net/forum?id=vh-0sUt8HlG, language=null, rfNumber=[49], rfOrder=58, authorNames=MEHTA S, RASTEGARI M, journalName=null, refType=null, unstructuredReference=MEHTA S, RASTEGARI M. MobileViT: light-weight, general-purpose, and mobile-friendly vision transformer[EB/OL]. [2024-12-28]. https://openreview.net/forum?id=vh-0sUt8HlG., articleTitle=MobileViT: light-weight, general-purpose, and mobile-friendly vision transformer, refAbstract=null)], funds=[Fund(id=1263550875470016991, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, awardId=62373151, language=EN, fundingSource=National Natural Science Foundation of China(62373151), fundOrder=null, country=null), Fund(id=1263550875717480935, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, awardId=62373151, language=CN, fundingSource=国家自然科学基金(62373151), fundOrder=null, country=null), Fund(id=1263550875948167660, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, awardId=F2023502010, language=EN, fundingSource=Natural Science Foundation of Hebei Province(F2023502010), fundOrder=null, country=null), Fund(id=1263550876229186034, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, awardId=F2023502010, language=CN, fundingSource=河北省自然科学基金(F2023502010), fundOrder=null, country=null), Fund(id=1263550876543758839, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, awardId=U21A20486, language=EN, fundingSource=Joint Funds of the National Natural Science Foundation of China(U21A20486), fundOrder=null, country=null), Fund(id=1263550878183731709, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, awardId=U21A20486, language=CN, fundingSource=国家自然科学基金联合基金项目重点支持项目(U21A20486), fundOrder=null, country=null), Fund(id=1263550878531858947, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, awardId=CXZZSS2025152, language=EN, fundingSource=Project for Cultivating the Innovative Ability of Full-time Postgraduate Students Studying in Hebei Province(CXZZSS2025152), fundOrder=null, country=null), Fund(id=1263550878821265929, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, awardId=CXZZSS2025152, language=CN, fundingSource=河北省在读研究生创新能力培养资助项目(CXZZSS2025152), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1263550824823795811, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, xref=null, ext=[AuthorCompanyExt(id=1263550824857350244, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Department of Automation, North China Electric Power University, Baoding Hebei 071003, China), AuthorCompanyExt(id=1263550824861544549, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, companyId=1263550824823795811, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华北电力大学自动化系, 河北 保定 071003)])], figs=[ArticleFig(id=1263550860626374959, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Fig. 1, caption=Display of the ten injury types ((a) Loss; (b) Glass breakage; (c) Glass scratches; (d) Mild deformation; (e) Moderate deformation; (f) Severe deformation; (g) Misalignment; (h) Tearing; (i) Body scuffing; (j) Body scratches), figureFileSmall=SiT+7IzpcHmLh1BfCHBu0A==, figureFileBig=158WvMvg/Jw5evM+BJALLA==, tableContent=null), ArticleFig(id=1263550861070971190, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=图1, caption=10类损伤类型展示((a) 丢失;(b) 玻璃破损;(c) 玻璃划痕;(d) 轻度变形;(e) 中度变形;(f) 重度变形;(g) 错位;(h) 撕裂;(i) 车身刮擦;(j) 车身划痕), figureFileSmall=SiT+7IzpcHmLh1BfCHBu0A==, figureFileBig=158WvMvg/Jw5evM+BJALLA==, tableContent=null), ArticleFig(id=1263550861494595901, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Fig. 2, caption=Structure of ResAWDNet model, figureFileSmall=b9ce4IveGJvLuGO+iiQh+A==, figureFileBig=nlni9StjCrlxvRXZExCalQ==, tableContent=null), ArticleFig(id=1263550861968552257, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=图2, caption=ResAWDNet模型结构图, figureFileSmall=b9ce4IveGJvLuGO+iiQh+A==, figureFileBig=nlni9StjCrlxvRXZExCalQ==, tableContent=null), ArticleFig(id=1263550862534783302, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Fig. 3, caption=Structure of weighted dynamic convolution, figureFileSmall=Mmc1/dXlhco6NabB3hZlTQ==, figureFileBig=WITEO/1/W58GbsjOlMVj0Q==, tableContent=null), ArticleFig(id=1263550862979379531, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=图3, caption=加权动态卷积结构图, figureFileSmall=Mmc1/dXlhco6NabB3hZlTQ==, figureFileBig=WITEO/1/W58GbsjOlMVj0Q==, tableContent=null), ArticleFig(id=1263550864900370768, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Fig. 4, caption=Structure of the dual attention module, figureFileSmall=o6kBG9j7Eu/CFyR3/SB3Ig==, figureFileBig=Qc6MFCDga4Bs86EqjEjG8A==, tableContent=null), ArticleFig(id=1263550865261080921, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=图4, caption=双重注意力模块结构图, figureFileSmall=o6kBG9j7Eu/CFyR3/SB3Ig==, figureFileBig=Qc6MFCDga4Bs86EqjEjG8A==, tableContent=null), ArticleFig(id=1263550865508544862, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Fig. 5, caption=Training process accuracy and loss curve, figureFileSmall=VGuujdOg0XnW4UJq8+o+Qg==, figureFileBig=wvaRr/6vrZPCnm2bHuuLwg==, tableContent=null), ArticleFig(id=1263550866011861346, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=图5, caption=训练过程准确率与损失曲线, figureFileSmall=VGuujdOg0XnW4UJq8+o+Qg==, figureFileBig=wvaRr/6vrZPCnm2bHuuLwg==, tableContent=null), ArticleFig(id=1263550866196410728, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Fig. 6, caption=Visualization comparison results before and after the introduction of the dual attention mechanism, figureFileSmall=ZCU7rONnOuX3i7RWdghJjg==, figureFileBig=pZIie36xwL/Vp7vfGI+hzQ==, tableContent=null), ArticleFig(id=1263550866548732268, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=图6, caption=引入双重注意力机制前后的可视化对比结果, figureFileSmall=ZCU7rONnOuX3i7RWdghJjg==, figureFileBig=pZIie36xwL/Vp7vfGI+hzQ==, tableContent=null), ArticleFig(id=1263550866968162675, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Fig. 7, caption=Classification results display ((a) Medium deformation; (b) Glass crack; (c) Missing;(d) Scratch; (e) Glass breakage; (f) Mild deformation; (g) Tearing; (h) Scratches; (i) Severe deformation; (j) Dislocation), figureFileSmall=cdTnaI5hT+a4cZY24cMKvA==, figureFileBig=hPhRa2GjL303OvH41vKJoQ==, tableContent=null), ArticleFig(id=1263550867249181049, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=图7, caption=分类结果展示((a) 中度变形;(b) 玻璃划痕;(c) 丢失;(d) 车身划痕;(e) 玻璃破损;(f) 轻度变形;(g) 撕裂;(h)车身刮擦;(i)重度变形;(j)错位), figureFileSmall=cdTnaI5hT+a4cZY24cMKvA==, figureFileBig=hPhRa2GjL303OvH41vKJoQ==, tableContent=null), ArticleFig(id=1263550869543465345, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Table 1, caption=

Data set composition

, figureFileSmall=null, figureFileBig=null, tableContent=
损伤类型 训练集/张 测试集/张 图片数量/张
错位 2 749 916 3 665
玻璃破损 913 304 1 217
玻璃裂痕 933 311 1 244
中度变形 1 493 497 1 990
轻微变形 1 488 496 1 984
丢失 2 683 894 3 577
车身划痕 4 336 1445 5 781
车身刮擦 4 116 1371 5 487
重度变形 1 455 485 1 940
撕裂 2 684 894 3 578
), ArticleFig(id=1263550870059364741, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=表1, caption=

数据集构成

, figureFileSmall=null, figureFileBig=null, tableContent=
损伤类型 训练集/张 测试集/张 图片数量/张
错位 2 749 916 3 665
玻璃破损 913 304 1 217
玻璃裂痕 933 311 1 244
中度变形 1 493 497 1 990
轻微变形 1 488 496 1 984
丢失 2 683 894 3 577
车身划痕 4 336 1445 5 781
车身刮擦 4 116 1371 5 487
重度变形 1 455 485 1 940
撕裂 2 684 894 3 578
), ArticleFig(id=1263550870424269194, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Table 2, caption=

Outcome of learning rate tuning

, figureFileSmall=null, figureFileBig=null, tableContent=
学习率 Acc_1%
0.01 51.67
0.005 53.58
0.001 62.45
0.000 5 69.50
0.000 1 73.79
0.000 05 73.19
0.000 01 72.81
), ArticleFig(id=1263550870722064784, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=表2, caption=

学习率调整结果

, figureFileSmall=null, figureFileBig=null, tableContent=
学习率 Acc_1%
0.01 51.67
0.005 53.58
0.001 62.45
0.000 5 69.50
0.000 1 73.79
0.000 05 73.19
0.000 01 72.81
), ArticleFig(id=1263550870927585685, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Table 3, caption=

Results of ablation experiments

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 Acc_1% Acc_5%
Baseline 71.88 97.24
Baseline+ WDConv 73.05 97.16
Baseline+DAM 72.97 97.14
ResAWDNet(本文模型) 73.79 97.68
), ArticleFig(id=1263550871103746457, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=表3, caption=

消融实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 Acc_1% Acc_5%
Baseline 71.88 97.24
Baseline+ WDConv 73.05 97.16
Baseline+DAM 72.97 97.14
ResAWDNet(本文模型) 73.79 97.68
), ArticleFig(id=1263550871317655971, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Table 4, caption=

Accuracy of classification of each injury type during ablation process

, figureFileSmall=null, figureFileBig=null, tableContent=
损伤类型 Baseline +DAM + WDConv ResAWDNet
错位 79.26 85.37 82.97 82.10
玻璃破损 76.64 78.62 80.92 77.96
玻璃裂痕 64.95 68.81 67.85 71.06
中度变形 25.75 47.89 23.94 30.99
轻度变形 51.81 46.17 53.23 49.80
丢失 73.60 73.60 78.19 80.09
车身划痕 90.73 89.34 88.86 92.25
车身刮擦 73.89 73.01 73.52 75.13
重度变形 52.99 63.30 70.72 61.44
撕裂 63.87 66.89 70.13 67.90
), ArticleFig(id=1263550871510593958, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=表4, caption=

消融过程中各损伤类型的分类准确率

, figureFileSmall=null, figureFileBig=null, tableContent=
损伤类型 Baseline +DAM + WDConv ResAWDNet
错位 79.26 85.37 82.97 82.10
玻璃破损 76.64 78.62 80.92 77.96
玻璃裂痕 64.95 68.81 67.85 71.06
中度变形 25.75 47.89 23.94 30.99
轻度变形 51.81 46.17 53.23 49.80
丢失 73.60 73.60 78.19 80.09
车身划痕 90.73 89.34 88.86 92.25
车身刮擦 73.89 73.01 73.52 75.13
重度变形 52.99 63.30 70.72 61.44
撕裂 63.87 66.89 70.13 67.90
), ArticleFig(id=1263550871657394602, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Table 5, caption=

Comparison of the effects of attention mechanisms

, figureFileSmall=null, figureFileBig=null, tableContent=
注意力机制 Acc_1% Acc_5%
Baseline 71.88 97.24
Baseline+SE[23] 72.32 97.48
Baseline+CBAM[25] 72.53 97.74
Baseline+EMA[27] 72.85 97.33
Baseline+EPSA[28] 72.61 97.36
Baseline+ECA[29] 72.93 97.62
Baseline+RGA[33] 72.49 97.35
Baseline+CPCA[34] 72.61 97.22
Baseline+DAM 72.97 97.14
), ArticleFig(id=1263550871883887023, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=表5, caption=

注意力机制效果对比

, figureFileSmall=null, figureFileBig=null, tableContent=
注意力机制 Acc_1% Acc_5%
Baseline 71.88 97.24
Baseline+SE[23] 72.32 97.48
Baseline+CBAM[25] 72.53 97.74
Baseline+EMA[27] 72.85 97.33
Baseline+EPSA[28] 72.61 97.36
Baseline+ECA[29] 72.93 97.62
Baseline+RGA[33] 72.49 97.35
Baseline+CPCA[34] 72.61 97.22
Baseline+DAM 72.97 97.14
), ArticleFig(id=1263550873653883313, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Table 6, caption=

Comparison of classification effects by injury type/%

, figureFileSmall=null, figureFileBig=null, tableContent=
损伤类型 Baseline ResAWDNet
Acc Pre Acc Pre
错位 79.26 83.75 82.10 81.56
玻璃破损 76.64 75.08 77.96 81.72
玻璃裂痕 64.95 73.49 71.06 74.16
中度变形 25.75 43.86 30.99 49.04
轻度变形 51.81 49.68 49.80 49.60
丢失 73.60 72.78 80.09 74.11
车身划痕 90.73 88.09 92.25 86.33
车身刮擦 73.89 68.72 75.13 71.23
重度变形 52.99 67.95 61.44 62.47
撕裂 63.87 68.63 67.90 77.03
), ArticleFig(id=1263550873934901689, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=表6, caption=

各损伤类型分类效果对比/%

, figureFileSmall=null, figureFileBig=null, tableContent=
损伤类型 Baseline ResAWDNet
Acc Pre Acc Pre
错位 79.26 83.75 82.10 81.56
玻璃破损 76.64 75.08 77.96 81.72
玻璃裂痕 64.95 73.49 71.06 74.16
中度变形 25.75 43.86 30.99 49.04
轻度变形 51.81 49.68 49.80 49.60
丢失 73.60 72.78 80.09 74.11
车身划痕 90.73 88.09 92.25 86.33
车身刮擦 73.89 68.72 75.13 71.23
重度变形 52.99 67.95 61.44 62.47
撕裂 63.87 68.63 67.90 77.03
), ArticleFig(id=1263550874308194753, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Table 7, caption=

Comparison with other models

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 Acc_1/% Acc_5/% Flops Params/M
AlexNet[36] 57.22 92.51 309.16 M 14.60
GoogleNet[37] 62.17 94.33 1.58 G 6.99
MobileNet[38] 58.08 94.02 327.55 M 3.50
ShuffleNet[39] 71.93 97.48 152.71 M 2.28
DenseNet[40] 72.72 97.11 2.90 G 7.98
EfficientNet[41] 69.80 96.97 412.83 M 5.29
RegNet[42] 72.77 97.65 207.35 M 2.32
EfficientNetv2[43] 71.97 97.01 2.89 G 21.46
FasterNet[44] 73.36 97.74 4.45 G 31.18
RepLKNet[45] 72.75 97.52 - 304.66
StarNet[46] 60.28 94.48 427.33 M 2.87
ResNet[17] 71.88 97.24 4.13 G 25.56
Vision Transformer[47] VIT-B16 64.59 95.97 16.88 G 103.03
VIT-B32 68.53 97.02 4.37 G 88.19
VIT-L16 72.32 97.90 59.69 G 304.12
VIT-L32 66.08 96.64 15.28 G 328.89
Swin Transformer[48] SwinT-T 72.76 97.60 4.37 G 28.27
SwinT-S 73.11 97.20 8.55 G 49.56
SwinT-B 72.90 97.65 23.57 G 109.07
MobileViT[49] 72.19 97.29 273.67 M 1.27
ResAWDNet 73.79 97.68 3.94 G 26.42
), ArticleFig(id=1263550874597601734, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=表7, caption=

与其他模型对比

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 Acc_1/% Acc_5/% Flops Params/M
AlexNet[36] 57.22 92.51 309.16 M 14.60
GoogleNet[37] 62.17 94.33 1.58 G 6.99
MobileNet[38] 58.08 94.02 327.55 M 3.50
ShuffleNet[39] 71.93 97.48 152.71 M 2.28
DenseNet[40] 72.72 97.11 2.90 G 7.98
EfficientNet[41] 69.80 96.97 412.83 M 5.29
RegNet[42] 72.77 97.65 207.35 M 2.32
EfficientNetv2[43] 71.97 97.01 2.89 G 21.46
FasterNet[44] 73.36 97.74 4.45 G 31.18
RepLKNet[45] 72.75 97.52 - 304.66
StarNet[46] 60.28 94.48 427.33 M 2.87
ResNet[17] 71.88 97.24 4.13 G 25.56
Vision Transformer[47] VIT-B16 64.59 95.97 16.88 G 103.03
VIT-B32 68.53 97.02 4.37 G 88.19
VIT-L16 72.32 97.90 59.69 G 304.12
VIT-L32 66.08 96.64 15.28 G 328.89
Swin Transformer[48] SwinT-T 72.76 97.60 4.37 G 28.27
SwinT-S 73.11 97.20 8.55 G 49.56
SwinT-B 72.90 97.65 23.57 G 109.07
MobileViT[49] 72.19 97.29 273.67 M 1.27
ResAWDNet 73.79 97.68 3.94 G 26.42
), ArticleFig(id=1263550874807316940, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=EN, label=Table 8, caption=

Comparison on the CarDD dataset

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 Acc_1/% Acc_5/%
ShuffleNet[39] 58.77 99.60
DenseNet[40] 59.09 99.84
FasterNet[44] 54.81 99.75
ResNet[17] 59.18 99.51
VIT-L16[47] 58.85 99.76
SwinT-S[48] 59.82 99.68
MobileViT[49] 60.15 99.78
ResAWDNet 60.43 99.68
), ArticleFig(id=1263550875130278356, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514357103715295, language=CN, label=表8, caption=

在CarDD数据集上的对比

, figureFileSmall=null, figureFileBig=null, tableContent=
模型 Acc_1/% Acc_5/%
ShuffleNet[39] 58.77 99.60
DenseNet[40] 59.09 99.84
FasterNet[44] 54.81 99.75
ResNet[17] 59.18 99.51
VIT-L16[47] 58.85 99.76
SwinT-S[48] 59.82 99.68
MobileViT[49] 60.15 99.78
ResAWDNet 60.43 99.68
)], attaches=null, journal=Journal(id=1263185177418154013, delFlag=0, nameCn=图学学报, nameEn=Journal of Graphics, nameHistory1=null, nameHistory2=null, issn=2095-302X, eissn=null, cn=10-1034/T, coden=null, periodic=1, language=CN, oaType=null, 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=o2S94GfCLjDYYJqsdQm5Mw==, journalPrice=null, startedYear=null, abbrevIsoEn=Journal of Graphics, journalRemark=null, publicationField=null, createdTime=1779096415144, updatedTime=1779097332356, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=J, firstLetterEn=J, subjectCode=Natural Sciences, subjectName=null, subjectCodeEn=Natural Sciences, subjectNameEn=null, picCn=o2S94GfCLjDYYJqsdQm5Mw==, picEn=f7IBTa1ZXKPoIhMUwL1eVw==, jcr=null, cjcr=null, exts=[JournalExt(id=1263189024693469540, language=CN, name=图学学报, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1779097332404, updatedTime=1779097332404, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.txxb.com.cn/Journalx_txxb/authorLogOn.action, submissionEditorUrl=http://www.txxb.com.cn/Journalx_txxb/editorLogOn.action, submissionReviewUrl=http://www.txxb.com.cn/Journalx_txxb/expertLogOn.action, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1263189024764772709, language=EN, name=Journal of Graphics, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1779097332421, updatedTime=1779097332421, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.txxb.com.cn/Journalx_txxb/authorLogOn.action, submissionEditorUrl=http://www.txxb.com.cn/Journalx_txxb/editorLogOn.action, submissionReviewUrl=http://www.txxb.com.cn/Journalx_txxb/expertLogOn.action, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1263187241531621409, websiteList=[Website(id=1263188159811207473, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1263187241531621409, 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/txxb/CN, language=CN, createTime=1779097126200, createBy=18614031015, updateTime=1779097561850, updateBy=18614031015, name=图学学报-中文, tplId=1146099689490845704, title=图学学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1263194793929204110, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=articleTextType, value=kx, createTime=1779098707897, updateTime=1779098707897, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793853706635, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=banner, value=null, createTime=1779098707879, updateTime=1779098707879, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793966952849, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=grayFlag, value=0, createTime=1779098707906, updateTime=1779098707906, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793824346506, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=logo, value=https://castjournals.cast.org.cn/joweb/txxb/CN/file/pic?fileId=lmN7m2FoR6FhgnIimGeZBg==, createTime=1779098707872, updateTime=1779098707872, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793992118675, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=minRunFlag, value=0, createTime=1779098707912, updateTime=1779098707912, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793891455373, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/txxb/CN/file/pic, createTime=1779098707888, updateTime=1779098707888, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793979535762, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=silenceFlag, value=0, createTime=1779098707909, updateTime=1779098707909, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793883066764, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1779098707886, updateTime=1779098707886, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793933398415, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=themeColor, value=null, createTime=1779098707898, updateTime=1779098707898, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793941787024, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=themeStyle, value=null, createTime=1779098707900, updateTime=1779098707900, creator=18614031015, updator=18614031015)]), Website(id=1263188160184500536, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1263187241531621409, 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/txxb/EN, language=EN, createTime=1779097126289, createBy=18614031015, updateTime=1779097557941, updateBy=18614031015, name=图学学报-英文, tplId=1146101810881728533, title=Journal of Graphics, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1263194821443838360, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=articleTextType, value=kx, createTime=1779098714457, updateTime=1779098714457, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821393506709, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=banner, value=null, createTime=1779098714445, updateTime=1779098714445, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194822144287131, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=grayFlag, value=0, createTime=1779098714624, updateTime=1779098714624, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821334786452, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=logo, value=https://castjournals.cast.org.cn/joweb/txxb/EN/file/pic?fileId=lmN7m2FoR6FhgnIimGeZBg==, createTime=1779098714431, updateTime=1779098714431, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194822316253597, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=minRunFlag, value=0, createTime=1779098714665, updateTime=1779098714665, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821414478231, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/txxb/EN/file/pic, createTime=1779098714450, updateTime=1779098714450, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194822249144732, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=silenceFlag, value=0, createTime=1779098714649, updateTime=1779098714649, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821401895318, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1779098714448, updateTime=1779098714448, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821716468121, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=themeColor, value=null, createTime=1779098714522, updateTime=1779098714522, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821850685850, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=themeStyle, value=null, createTime=1779098714554, updateTime=1779098714554, creator=18614031015, updator=18614031015)])], journalTitle=图学学报, weixinUrl=null, journalUrl=http://www.txxb.com.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Journal of Graphics, journalPhotoCn=o2S94GfCLjDYYJqsdQm5Mw==, journalPhotoEn=f7IBTa1ZXKPoIhMUwL1eVw==, journalFirstLetter=J, 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=, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/txxb/CN/10.11996/JG.j.2095-302X.2026010017, detailUrlEn=https://castjournals.cast.org.cn/joweb/txxb/EN/10.11996/JG.j.2095-302X.2026010017, pdfUrlCn=https://castjournals.cast.org.cn/joweb/txxb/CN/PDF/10.11996/JG.j.2095-302X.2026010017, pdfUrlEn=https://castjournals.cast.org.cn/joweb/txxb/EN/PDF/10.11996/JG.j.2095-302X.2026010017, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
融合双重注意力与加权动态卷积的车辆损伤分类模型
收藏切换
PDF下载
翟永杰 , 王紫萱 , 张祯琪 , 周迅琪 , 王乾铭
图学学报 | 图像处理与计算机视觉 2026,47(1): 17-28
收起
收藏切换
图学学报 | 图像处理与计算机视觉 2026, 47(1): 17-28
融合双重注意力与加权动态卷积的车辆损伤分类模型
全屏
翟永杰, 王紫萱, 张祯琪, 周迅琪, 王乾铭
作者信息
  • 华北电力大学自动化系, 河北 保定 071003

通讯作者:

王乾铭,E-mail:
A vehicle damage classification model incorporating dual attention and weighted dynamic convolution
Yongjie ZHAI, Zixuan WANG, Zhenqi ZHANG, Xunqi ZHOU, Qianming WANG
Affiliations
  • Department of Automation, North China Electric Power University, Baoding Hebei 071003, China
出版时间: 2026-02-28 doi: 10.11996/JG.j.2095-302X.2026010017
文章导航
收藏切换

针对车险理赔客户上传的车辆损伤图像中存在损伤类型形态相似、分类困难的问题,提出了一种适用于车辆损伤分类的模型ResAWDNet。首先,为有效增强模型对损伤特征的提取能力,使用加权动态卷积代替原有的下采样操作,依据输入特征动态调整卷积核权重,提高模型对不同尺度和方向特征的适应性,从而更准确地捕捉损伤的细微差异。其次,为了使模型关注图像中的显著性判别区域和特征通道,在主干网络的卷积层后嵌入了双重注意力机制,同时学习空间和通道维度上的重要权重,提升模型对关键信息的捕捉能力,进一步提升模型在损伤分类任务中的决策准确性。最后,基于真实事故案例的车辆损伤图片数据集进行实验验证。实验结果表明,ResAWDNet模型在车辆损伤分类任务中切实可行且优势显著,整体分类准确率达到73.79%。与基线模型相比,ResAWDNet在多类损伤类型的分类上均展现出更高的准确率,有力地证明了该模型的有效性。

智能定损  /  图像分类  /  深度学习  /  注意力机制  /  动态卷积

To address the challenges of morphological similarity and the resulting difficulty in classifying vehicle damage images uploaded by clients for auto insurance claims, a model named ResAWDNet was proposed for vehicle damage classification. Firstly, to effectively augment the model’s capacity for extracting damage features, the traditional down sampling operation was replaced with weighted dynamic convolution. This approach dynamically adjusted the weights of convolutional kernels based on the input features, thereby enhancing the model’s adaptability to features of varying scales and orientations. As a result, it enabled more precise capture of the subtle differences in vehicle damage. Secondly, to ensure that the model could concentrate on the salient discriminative regions and feature channels within the images, a dual attention mechanism was embedded after the convolutional layers of the backbone network. This mechanism concurrently learned the important weights in both spatial and channel dimensions, significantly enhancing the model’s ability to capture crucial information. Consequently, it further enhanced the decision-making accuracy of the model in the task of vehicle damage classification. Finally, experimental validation was conducted based on a dataset of vehicle damage images sourced from real accident cases. The experimental results demonstrated that the ResAWDNet model was feasible and offered significant advantages for vehicle damage classification tasks, achieving an accuracy rate of 73.79%. Compared with baseline models, ResAWDNet achieved higher accuracy in classifying multiple types of damages, robustly validating the effectiveness of the proposed model.

intelligent damage assessment  /  image classification  /  deep learning  /  attention mechanism  /  dynamic convolution
翟永杰, 王紫萱, 张祯琪, 周迅琪, 王乾铭. 融合双重注意力与加权动态卷积的车辆损伤分类模型. 图学学报, 2026 , 47 (1) : 17 -28 . DOI: 10.11996/JG.j.2095-302X.2026010017
Yongjie ZHAI, Zixuan WANG, Zhenqi ZHANG, Xunqi ZHOU, Qianming WANG. A vehicle damage classification model incorporating dual attention and weighted dynamic convolution[J]. Journal of Graphics, 2026 , 47 (1) : 17 -28 . DOI: 10.11996/JG.j.2095-302X.2026010017
中华人民共和国国家发展和改革委员会于2020年发布“数字化转型伙伴行动”倡议[1],旨在推动各行业的数字化转型,提升企业的智能化水平。对于保险公司而言,结合现代智能技术实现车辆智能定损以进行车险智慧化转型是实现业务提升的重要途径,也是提升保险公司运营效率、客户满意度的有效手段。车辆智能定损技术[2]利用深度学习算法,采用人工拍摄方式获取车辆损伤外观图像,通过分析大量汽车相关数据和损伤图像判定汽车的损坏情况和损伤类型,进而确定受损类型以及评估部件受损程度,给出赔付方案。因此,基于计算机视觉的车辆智能定损技术能够有效降低保险定损时投入的人力和物力成本,提高车辆定损效率。
智能定损中常见的损伤可分为车身剐蹭、车身划痕、车身变形、玻璃破损、玻璃划痕、错位等多种类别,如图1所示。由于车辆的外观部件形态多样,且损伤发生的情况、角度和场景均不同,导致同一类别的损伤在视觉上表现差异较大,难以分辨。同时损伤图片存在模糊、质量不佳的问题,为损伤辨别增加了难度。因此需要提出针对性的模型能够迅速且精准地对车辆损伤图像进行分类。
近年来,深度学习技术在损伤分类领域已经取得了诸多进展和成就。如武兵和田莹[3]提出的YOLOv8-RDD模型应用于道路损伤检测;LIU等[4]开发了基于Transformer结合分块与掩码机制的模型用于工业气缸套缺陷检测;PARK等[5]提出了一种卷积神经网络成像分析,识别检测零件表面缺陷;WU等[6]针对小样本下的钢表面缺陷分类问题,优化了特征提取网络,为实际工业生产中的小样本损伤分类提供了新思路;LIU等[7]提出了一种基于多尺度特征融合的深度学习模型,用于复杂工业零部件的表面损伤分类。然而,上述工作大多针对金属缺陷、表面划痕等单一损伤类型,车辆损伤类型多样,形态多变,已有的算法在车辆损伤分类任务中难以取得理想表现,针对车辆损伤分类的具体研究仍处于相对初级的探索阶段。
本文结合保险公司实际车险案例图片构建车辆损伤数据集,基于ResNet50模型,针对车辆损伤分类任务进行了深入研究和改进。通过嵌入双重注意力、使用加权动态卷积代替原有的下采样,提出一种适用于车辆损伤分类的模型。本文主要贡献如下:
1) 以ResNet50模型作为基础框架,提出了适用于车辆损伤的分类模型ResAWDNet,实现了车辆损伤特征的高效识别与分类。
2) 采用加权动态卷积实现下采样操作,增强模型特征提取能力,提高模型对不同特征的适应性;提出双重注意力模块并嵌入卷积层后,提高模型对图像中关键信息的捕捉能力,提升模型的分类性能与准确性。
3) 基于真实事故案例采集图像,验证了本文提出的ResAWDNet模型在车辆损伤分类任务中的可行性和优越性。
图像分类是计算机视觉领域的重要研究方向之一,其旨在利用图像的视觉特征自动将图片划分到某一类别[8]。常见的图像分类算法包括传统图像分类算法和基于深度学习的图像分类算法[9]。传统算法需人工设计特征提取器通过SVM (Support Vector Machine)、随机森林等分类器进行决策[10],例如贺敏雪等[11]融合车标对称性特征通过CRC(Collaborative Representation Based Classification)分类器识别;ANANDA和PUTRI[12]利用K-最近邻算法对车辆损伤特征进行分类匹配,结合案例推理借鉴过往案例经验,有效提高了车损检测的准确性与效率。然而,人工设计的特征在捕捉复杂特征形态方面存在局限性,难以契合实际业务中车辆损伤多样化的需求。
随着深度学习技术的迅猛发展,基于深度学习的车辆损伤图像分类算法在近年来逐渐成为研究热点。MISHRA等[13]基于YOLOv8和ResNet50构建的MD R-CNN(Multi-Detection Region- Convolutional Neural Networks)模型,通过多尺度特征融合提升车辆损伤分类性能,但对不良光照下的损伤识别效果有限;PENG等[14]提出了多视角融合与对齐的车辆损伤检测模型MVA-CDD(Multi- View Car Damage Detection Model),但对小尺度损伤的敏感度较低;SHUBHAM和BANERJEE[15]运用卷积神经网络(Convolutional Neural Networks, CNN)提取车辆损伤图像特征,并结合SVM进行分类,有效提高了损伤识别精度,但对复杂光照、多角度拍摄的损伤图像鲁棒性欠佳。王心旷[16]提出基于可变形卷积网络(Deformable Convolution Networks, DCN)改进的DCN+算法,通过集成多尺度训练策略、采用混合焦点损失函数等手段,提升了小尺度车损目标的识别准确率,但在刮擦、凹陷等复杂类别的识别精度方面仍有待进一步提高。
尽管上述研究通过优化网络架构、改进特征提取机制以及创新损失函数设计等手段,在模型精度提升、计算效率优化等方面取得一定进展,但在车险理赔的特殊应用场景中仍面临着形态相似损伤特征区分困难、模型泛化能力弱的问题。为此,本文提出了一种适用于车辆损伤的分类网络,旨在为车辆定损提供更准确、高效的解决方案。
下采样(又名降采样),是指通过一定的算法或操作,将数据的空间分辨率降低、减少数据量的过程。在深度学习中,较低层次的特征通常对应于局部的细节信息,而较高层次的特征则更加抽象,反映了全局结构和语义信息。通过下采样,模型可以逐渐从局部细节中抽象出更具概括性的特征,从而更好地适应不同的任务需求。如在图像分类任务中,ResNet[17]通过步幅卷积实现下采样操作,减少特征图的尺寸,从而降低计算复杂度;在目标检测任务中,Faster R-CNN(Faster Region-Convolutional Neural Networks)[18]通过最大池化和步幅卷积实现下采样,用于生成不同尺度的特征图,以便检测不同大小的目标;在语义分割任务中,DeepLab[19]系列通过空洞卷积和步幅卷积实现下采样,以便更好地捕捉图像的上下文信息;在生成对抗网络(Generative Adversarial Networks, GAN)中,Progressive GAN[20]通过步幅卷积实现下采样,生成不同分辨率的图像。
然而,目前常用池化操作和步幅卷积的下采样方法存在一定局限性。一方面其采样特征难以适配多样化图像内容,面对复杂多变的图像场景,缺乏灵活性与针对性,导致特征提取效果不佳。另一方面,在处理过程中,全局性信息利用不足,模型无法充分捕捉图像全局结构与语义关联,限制了对图像整体理解,影响任务精度与性能[21]。针对这一问题,本文提出了加权动态卷积下采样方式,通过自适应调整卷积核权重,实现对不同图像区域的有针对性采样,提高模型的泛化性。
注意力机制能够帮助模型聚焦于图像中与任务相关的区域,改善模型对于关键信息的处理能力,从而提升模型的性能[22]。如HU等[23]关注通道之间的关系提出的SE(Squeeze-and-Excitation Networks); JADERBERG等[24]基于空间关系注意力提出的STN(Spatial Transformer Networks)和WOO等[25]提出融合空间注意力机制和通道注意力机制的CBAM(Convolutional Block Attention Module)等注意力机制,验证了注意力机制增强模型表征能力的有效性。
近年来,尽管注意力机制在深度学习领域快速发展,但在车辆损伤图像分类任务中表现欠佳。SeerAttention[26]通过可学习的门自适应选择重要区块平衡准确性和速度,但主要针对面向语言任务,难以捕捉车辆损伤的非规则分布。EMA(Efficient Multi-scale Attention)[27]依赖时序特征加权,难以在静态图像中动态捕捉局部关键区域。EPSA(Efficient Pyramid Squeeze Attention)[28]多尺度空间聚合但忽略了通道交互,ECA(Efficient Channel Attention)[29]纯通道增强缺乏空间敏感性,复杂损伤形态的表征能力受限。综上,由于任务目标、数据特性及模型需求存在差异,现有的注意力机制难以直接应用到车辆损伤分类任务中。
针对车险理赔客户上传的车辆损伤图像中存在损伤类型形态相似、分类困难的问题,本文提出了一种适用于车辆损伤分类的模型ResAWDNet。以ResNet50作为基础框架,并引入加权动态卷积(Weighted Dynamic Convolution, WDConv)与双重注意力机制(Dual Attention Mechanism, DAM)来优化车辆损伤分类模型。模型结构如图2所示。
图像输入模型后,首先会经过Stage0和Stage1的初步处理。从Stage2开始,模型采用加权动态卷积替代ResNet50网络Stage2~Stage4中原有的下采样操作。接着依次经过卷积操作,完成特征提取与变换,最终输出特征图。Stage2~Stage4均采用这样的加权动态卷积处理,各阶段处理后的特征图作为后续阶段的输入,实现了特征的逐步丰富与细化。
经Stage4处理后的特征图,作为双重注意力模块的输入。在此模块中,通过通道注意力与空间注意力依次处理,引导模型重点关注图像中的关键区域以及具有重要信息的特征通道,提高模型对于损伤部位的敏感程度,进一步提升整体分类性能。
加权动态卷积下采样层的核心作用是动态地调整卷积核权重,利用调整后的权重对经过填充处理的输入特征图进行分组卷积操作。相比于传统的固定卷积核,动态卷积能够根据输入数据的特征动态调整卷积核的权重,有助于提取更具有代表性的特征。在下采样的过程中,模型需保留更多有用的信息,以减少特征信息的丢失。本文设计并采用加权动态卷积进行下采样操作,使得下采样操作更加灵活,可以根据输入数据的不同,动态地调整卷积核的权重,从而更好地适应不同的任务需求。加权动态卷积结构如图3所示。
输入的特征图$F$首先通过计算得卷积核权重${w}_{\text{1}}$,该过程采用分组卷积的形式,即将输入通道分成若干组,每组独立进行卷积操作,即
${w}_{\text{1}}=\text{Conv}(F),BN(F)$
式中:Conv表示卷积操作;BN表示批量归一化。
随后将权重${w}_{\text{1}}$重新塑形,并在通道维度上重复,使其与输入特征图的通道数匹配。塑形后的权重为
$w_{1}^{\prime} \in R^{\left(B, G, 1, K^{2}, H, W\right)}$
式中:B表示一次处理的样本大小;G表示分组数量;1为维度;K表示卷积核大小;HW分别表示输出特征图的高度和宽度。
再通过计算得到权重${w}_{2}$,重新塑形后记为$w_{2}^{\prime}$。将权重$w_{1}^{\prime}$$w_{2}^{\prime}$相乘,以调整每个通道的权重,输出结果记为${w}_{3}$。此过程可以表示为
${w}_{2}=Conv[Conv(\mathrm{AvgPool}(F))]$
$w_{2}^{\prime} \in R^{(B, G, C / G, 1,1,1)}$
$w_{3}=w_{1}^{\prime} \times w_{2}^{\prime}$
式中:AvgPool表示平均池化;B表示一次处理的样本大小;G表示分组数量;C表示输入张量的通道数。
文献[30]提到,低通滤波器能够减少图像中噪声对特征的干扰。因此将得到的权重${w}_{3}$经过Softmax操作,约束权重为正并求和为1,使得生成的权重是低通的,从而减弱特征噪声的干扰。随后重新调整${w}_{3}$的形状,以便在后续的分组卷积操作中能够正确地进行计算,即
$w_{3}^{\prime}=\operatorname{reshape}\left[\operatorname{Softmax}\left(w_{3}\right)\right]$
对输入特征图进行反射填充,并展开生成滑动窗口的特征块。最后,将得到的权重${{w}^{\prime }}_{3}$与展开后的特征块相乘,并对卷积核维度进行求和,生成最终的输出特征图,即
$pa{d}_{F}=pad(F)$
$\text { pad }_{F}^{\prime}=\text { reshape }\left[\text { unfold }\left(\text { pad }_{F}\right)\right]$
$\boldsymbol{F}_{\mathrm{out}}=\operatorname{reshape}\left[\operatorname{sum}\left(w_{3}^{\prime} \times \operatorname{pad}_{F}^{\prime}\right)\right]$
加权动态卷积模块将自适应加权的思想与分组卷积相结合,充分利用局部信息和全局信息调整卷积核权重,使模型在轻量化的同时能够加强表达能力,整体算法流程如下:
在车辆损伤分类任务中,所使用的数据集具有车辆损伤形态多样、相似性高的显著特点,仅依据单一维度的特征识别方法,极易造成分类错误。因此,本文提出了一种双重注意力模块DAM (Dual Attention Modu),采用一维卷积高效实现局部跨通道交互,使模块能够整合不同通道中形状、光影和纹理等信息,实现对车辆损伤特征的全面、精准提取。同时,DAM模块借鉴了CBAM的设计思想,将通道注意力与空间注意力进行有机结合,提升模型对图像特征的表征能力,增强模型决策的准确性。双重注意力模块的结构如图4所示。
输入的特征图$F$经过全局平均池化,压缩其空间维度,获取每个通道的全局信息。这一步骤将特征图从$\text{(}N\text{,}C\text{,}W\text{,}H\text{)}$大小转换为$\text{(}N\text{,}C\text{,1,1)}$,其中N为批次大小,C为通道数,HW分别为特征图的高度和宽度。此过程可以表示为
${Z}_{c}=1/\left(H\times W\right){\displaystyle \sum _{i=1}^{H}{\displaystyle \sum _{j=1}^{W}{x}_{c}}}(i,j)$
式中:${Z}_{c}$表示第$c$个通道的全局平均池化结果;${x}_{c}(i,j)$表示第$c$个通道在位置$(i,j)$的像素值。
随后,确定一维卷积核的大小$k$,以确定局部跨通道交互的覆盖范围。卷积核大小与通道数成正相关,通道数越大则选取的卷积核越大。二者关系可表示为
$k=\varphi (c)$
使用计算得到的一维卷积核对全局平均池化后的特征图进行一维卷积操作,并通过Sigmoid激活函数生成每个通道的权重,即
${w}_{c}=\sigma \cdot \text{Conv1}(z,k)$
式中:$\text{Conv}1\text{(}z,k\text{)}$表示卷积操作;$z$表示全局平均池化后的特征图;$k$表示一维卷积核;$\sigma $表示Sigmoid激活函数;${w}_{c}$表示第$c$个通道的权重。
再将生成的通道权重与原始输入特征图F进行对应元素相乘,得到输出特征图F1。此过程可以表示为
$\tilde{f}={f}_{c}\cdot {w}_{c}$
$\left({f}_{1},{f}_{2},{f}_{3},\cdot \cdot \cdot,{f}_{c}\right)\in {F}_{1}$
式中:$\tilde{f}$表示加权后的第$c$个通道的特征图;${f}_{c}$表示原始输入特征图的第$c$个通道。
第一阶段得到的特征图$({\mathrm{f}}_{1},{\mathrm{f}}_{2},{\mathrm{f}}_{3},\cdot \cdot \cdot,{\mathrm{f}}_{\mathrm{c}})\in {\mathrm{F}}_{1}$在空间维度进行池化,获取特征图在空间维度上的平均信息和最显著信息,并将得到的2个二维特征图进行拼接即
$ou{t}_{1}=\text{Concat}[\text{Maxpool}({F}_{1}),\text{AvgPool}({F}_{1})]$
通过一个卷积层进行特征融合和降维,将卷积层的输出通过Sigmoid激活函数,得到归一化的空间注意力权重。权重与特征图F1相乘得到最终的输出特征为
${w}_{2}=\sigma \cdot \text{Conv}2(ou{t}_{1})$
${F}_{\text{out}}={F}_{1}\cdot {w}_{2}$
通过融合通道注意力与空间注意力机制,DAM模块能够增强模型对图像重要特征的捕捉能力。在输出层前加入注意力机制可以使网络更加聚焦于最终的决策过程,对前面层提取的特征进行最后的筛选和加权,确保网络的输出更加准确地反映输入图像的关键特征与任务目标的关系[31]。因此本文在输出层前引入双重注意力模块,使得重要的特征得到更多关注。
本文使用车辆损伤分类数据集对本文模型进行评估,车辆损伤分类数据集由保险公司提供,该数据集来源于其在车险理赔业务中的实际车辆损伤勘查记录,包含车身、前窗和侧窗等多类车辆部位,涵盖错位、玻璃破损、玻璃裂痕、中度变形、轻微变形、丢失、车身划痕、车身刮擦、重度变形和撕裂共10类损伤类型。总计30 463张图片,其中训练集与测试集样本图像比例为3∶1,分别为22 850和7 613张。数据集各类图片数量见表1
为进一步评估本模型的性能,本文还选用公开的车辆损伤数据集CarDD[32]进行实验,该数据集涵盖凹痕、划痕、裂纹、玻璃破碎、轮胎漏气和车灯损坏6种常见车辆损伤类型,适用于多种计算机视觉任务。
本文所用操作系统为Ubuntu 18.04 LTS,GPU为Geforce RTX 3090显卡,CUDA版本为12.5,以Python3.8为程序开发语言。采用随机梯度下降法(Stochastic Gradient Descent,SGD)作为优化模型的方法,输入图像尺寸设为224×224×3像素,批次大小设为8,经实验确定学习率设为0.000 1,迭代次数设为100。学习率调整结果见表2、训练过程准确率与损失曲线如图5所示。
以每一类别的准确率Accuracy、每一类别的精确率Precision、整体的Top-1准确率和Top-5准确率作为车辆损伤分类的评价标准。其中Top-1准确率是指排名第一的类别与实际结果相符的准确率,简称为准确率;Top-5准确率是指排名前5的类别包含实际结果的准确率。两者大小均介于[0,1]之间,数值越大,表示分类效果越好。每一类别的准确率及精确率计算式分别为
$Accuracy=\frac{TP+TN}{TP+TN+FP+FN}$
$Precision=\frac{TP}{TP+FP}$
式中:TP (True Positive)表示正样本被正确识别的数量;FP (False Positive)表示误报的负样本数量;TN (True Negitive)表示负样本被正确识别的数量;FN (False Negitive)表示漏报的正样本数量。
本文以ResNet模型为基线模型实现了所提出模型ResAWDNet,将各模块在保险公司真实应用场景下的车辆损伤分类数据集上进行实验并测试了2种定量指标,实验结果见表3。Baseline代表基础模型,Baseline+WDConv代表引入加权动态卷积下采样的模型,Baseline+DAM代表加入双重注意力的模型,ResAWDNet代表结合双重注意力和加权动态卷积下采样的模型。
表3可知,基线模型Top-1准确率、Top-5准确率分别为71.88%和97.24%,而分别引入WDConv和DAM下采样使Top-1准确率分别提升了1.17%和1.09%。两者结合构成ResAWDNet使Top-1准确率达到73.79%,相较基线模型提升1.91%,超越单一模块的改进幅度,体现了模块设计的有效性。
表4展示了消融过程中各损伤类型的分类准确率。由表中结果可以看出引入DAM和WDConv下采样模块对大多数损伤类型分类准确率均有提升,其中引入DAM机制对错位、中度变形2类损伤影响较大,采用WDConv下采样对玻璃破损、撕裂等损伤影响较大。尽管消融过程中部分类别的准确率存在一定的波动,但从整体实验结果来看,二者结合使模型在损伤分类中展现出更强的能力,验证ResAWDNet的融合策略是有效的。
为验证DAM的有效性,本文进行了一系列对比实验。在保证超参数设置相同的情况下,选取ResNet50作为基础模型框架,对比加入SE[23]通道注意力、CBAM[25]卷积注意力等注意力机制与加入本文所提出的DAM机制后的性能表现,具体内容见表5。由表中结果可知,相较于未引入任何注意力机制的基线ResNet50模型,应用SE[23],CBAM[25]和ECA[29]等注意力机制均能在一定程度上提升模型在图像分类任务上的准确率。然而,在相同的实验条件下,本文提出的加入DAM机制的模型表现尤为突出,相较于基线模型分类准确率提高了1.09%,与近年提出的EMA和EPSA等注意力相比也占优势。
图6展示了在基线模型中引入DAM机制前后的可视化对比结果,通过热力图[35]呈现了模型在处理图像数据时对不同区域的关注度。在未加入注意力机制的情况下,基线模型的注意力分布较为分散,未能有效聚焦于图像中的关键损伤部位,加入DAM机制后,模型更加专注于图像中的损伤区域,有效减少了无关信息的干扰,显著提升了模型对关键信息的捕捉能力。如在变形损伤中,加入DAM机制后热力图的高热度区域与损伤位置贴合更为准确,能够准确覆盖变形区域,表明其能更有效地聚焦关键损伤特征。对比基线模型,加入DAM机制后热力图中高热度区域集中在玻璃实际破损或裂痕周围,即使在反光影响下,也能有效聚焦于损伤本身,充分体现出DAM机制在反光等复杂光照条件下对损伤检测的优势。
基于车辆损伤数据集,对基线模型和本文提出的ResAWDNet模型进行了对比。从表6结果可以看出,ResAWDNet在多数损伤类型上的分类准确率显著高于基线模型,其中,“玻璃裂痕”分类准确率由64.95%提升至71.06%,相较于基线模型提升了6.11%;“重度变形”分类准确率由52.99%提升至61.44%,提升了8.45%。尽管部分类别的分类准确率仍有提升空间,但从整体角度看,ResAWDNet在多数类别上实现了准确率的提升。
为验证本文所提出模型的优越性,将数据集分别在AlexNet和GoogleNet等经典的卷积神经分类网络模型、RepLKNet、StartNet等前沿的网络模型和ViT、SwinT等Transformer模型中进行训练和测试,结果见表7。通过对比各模型的实验结果,可以明显看出,与其他经典的CNN模型相比,本文提出的结合DAM和WDConv下采样的ResAWDNet模型在分类任务中展现出了显著的优势。具体来说,ResAWDNet在Top-1和Top-5的准确率分别达到了73.79%和97.68%,均优于如AlexNet(57.22%)、GoogleNet(62.17%)、MobileNet (58.08%)等经典模型,以及MobileViT(72.19%)、DenseNet(72.72%)、RepLKNet(72.75%)等前沿模型,充分体现了ResAWDNet模型在车辆损伤分类任务中的有效性和先进性。
为进一步评估本模型在损伤分类任务中的泛化性,选用公开的车辆损伤数据集CarDD进行对比实验。表8展示了本方法与当前主流图像分类算法在CarDD数据集上的对比结果,从中可看出,本模型在分类表现上更为优异。
为检验ResAWDNet模型的分类准确性,对其分类结果进行展示,通过输入不同的图片,预测其损伤类型。图7展示了基线模型和本文ResAWDNet模型的分类结果。与基线模型相比,ResAWDNet模型展现出了更为出色的分类能力,在处理相似或模糊损伤特征时也表现出了较高的准确率,充分验证了ResAWDNet模型在损伤类型分类任务上的优越性和可靠性。
本文研究致力于提升车辆损伤分类算法的精度。本文提出了DAM机制与WDConv模块,并将二者相结合,以ResNet50作为基础框架构建了ResAWDNet模型,实现了在车辆损伤分类任务中的性能提升。WDConv模块能够根据不同输入动态调整卷积核,优化特征提取过程,有效减轻数据集中低质量图像对分类性能的不利影响。DAM机制能够同时捕捉通道和空间维度的关键信息,使模型在处理车辆损伤的相似图像数据时更加高效和准确。在车辆损伤数据集上的实验结果表明,ResAWDNet模型在车辆损伤分类任务中表现出色。与基线模型相比,ResAWDNet准确率提升了1.91%,达到了73.79%。在各类损伤类型的分类上,ResAWDNet也普遍表现出更高的准确率,特别是在错位、玻璃裂痕和丢失等类型上,准确率提升尤为显著。
本文通过提出DAM机制与WDConv模块,并将其有效结合构建ResAWDNet模型,针对车辆损伤分类任务中的图像损伤特征形态相似性高、分类难度大的问题进行了深入研究,为车辆多形态部件损伤缺陷精确识别提供了有效的技术方案。然而,该方法也存在一定的局限性,对相互遮挡的损伤区域特征提取能力有限,分类准确率有待提升。在未来的工作中,将探索更鲁棒的特征提取与解耦机制,与多目标检测算法结合,部署到车辆理赔现场实际运行环境中,从客户端上传待定损图像到对车辆部件状态研判,实现端到端部署和运行。
  • 国家自然科学基金(62373151)
  • 河北省自然科学基金(F2023502010)
  • 国家自然科学基金联合基金项目重点支持项目(U21A20486)
  • 河北省在读研究生创新能力培养资助项目(CXZZSS2025152)
参考文献 引证文献
排序方式:
[1]
赵子豪, 申颖, 李薇. 基于图像识别的车辆智能定损应用研究[J]. 保险职业学院学报, 2019, 33(3): 73-77.
ZHAO Z H, SHEN Y, LI W. Application and value research about apps of vehicle survey and loss assessment based on image recognition[J]. Journal of Insurance Professional College, 2019, 33(3): 73-77 (in Chinese).
[2]
翟永杰, 李佳蔚, 陈年昊, . 融合改进Transformer的车辆部件检测方法[J]. 图学学报, 2024, 45(5): 930-940.
ZHAI Y J, LI J W, CHEN N H, et al. The vehicle parts detection method enhanced with Transformer integration[J]. Journal of Graphics, 2024, 45(5): 930-940 (in Chinese).
[3]
武兵, 田莹. 基于注意力机制的多尺度道路损伤检测算法研究[J]. 图学学报, 2024, 45(4): 770-778.
WU B, TIAN Y. Research on multi-scale road damage detection algorithm based on attention mechanism[J]. Journal of Graphics, 2024, 45(4): 770-778 (in Chinese).
[4]
LIU Q, HUANG X H, SHAO X Y, et al. Industrial cylinder liner defect detection using a transformer with a block division and mask mechanism[J]. Scientific Reports, 2022, 12(1): 10689.
[5]
PARK J K, KWON B K, PARK J H, et al. Machine learning-based imaging system for surface defect inspection[J]. International Journal of Precision Engineering and Manufacturing-Green Technology, 2016, 3(3): 303-310.
[6]
WU S Q, ZHAO S Y, ZHANG Q Q, et al. Steel surface defect classification based on small sample learning[J]. Applied Sciences, 2021, 11(23): 11459.
[7]
LIU W, QIU J, WANG Y, et al. Multiscale feature fusion convolutional neural network for surface damage detection in retired steel shafts[J]. Journal of Computing and Information Science in Engineering, 2024, 24(4): 041005..
[8]
王瑞芳. 基于字典学习的图像分类算法研究[D]. 重庆: 重庆邮电大学, 2020.
WANG R F. Research on image classification algorithm based on dictionary learning[D]. Chongqing: Chongqing University of Posts and Telecommunications, 2020 (in Chinese).
[9]
张鹏飞, 石志良, 李晓垚, . 基于深度学习的主轴承盖分类识别算法[J]. 图学学报, 2021, 42(4): 572-580.
ZHANG P F, SHI Z L, LI X Y, et al. Classification algorithm of main bearing cap based on deep learning[J]. Journal of Graphics, 2021, 42(4): 572-580 (in Chinese).
[10]
董潇. 卷积神经网络的图像分类优化算法研究[D]. 淮南: 安徽理工大学, 2020.
DONG X. Research on image classification optimization algorithm of convolutional neural network[D]. Huainan: Anhui University of Science & Technology, 2020 (in Chinese).
[11]
贺敏雪, 余烨, 程茹秋. 特征增强策略驱动的车标识别[J]. 中国图象图形学报, 2021, 26(5): 1030-1040.
HE M X, YU Y, CHENG R Q. Vehicle logo recognition method based on feature enhancement[J]. Journal of Image and Graphics, 2021, 26(5): 1030-1040 (in Chinese).
[12]
ANANDA B, PUTRI R A. K-nearest neighbor algorithm and case base reasoning on xenia car damage detection expert system[J]. Journal of Computer Networks, Architecture and High Performance Computing, 2024, 6(2): 633-646.
[13]
MISHRA S, KAMAL D, SENTHIL KUMAR K. Vehicle damage identification using deep learning techniques[C]// 2024 IEEE International Students' Conference on Electrical, Electronics and Computer Science. New York: IEEE Press, 2024: 1-6.
[14]
PENG J B, DONG S B, YUAN H, et al. Car damage detection based on multi-view fusion and alignment: dataset and method[J]. IEEE Transactions on Intelligent Transportation Systems, 2025, 26(4): 4717-4730.
[15]
SHUBHAM, BANERJEE D. Robust car damage identification through CNN and SVM techniques[C]// The 4th International Conference on Technological Advancements in Computational Sciences. New York: IEEE Press, 2024: 101-107.
[16]
王心旷. 基于深度学习的车辆外观损伤识别及其图像生成方法研究[D]. 合肥: 中国科学技术大学, 2024.
WANG X K. Research on car exterior damage recognition and image generation based on deep learning[D]. Hefei: University of Science and Technology of China, 2024 (in Chinese).
[17]
HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016: 770-778.
[18]
REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.
[19]
CHEN L C, PAPANDREOU G, KOKKINOS I, et al. Semantic image segmentation with deep convolutional nets and fully connected CRFs[EB/OL]. [2024-12-28]. https://arxiv.org/abs/1412.7062. https://arxiv.org/abs/1412.7062.
[20]
KARRAS T, AILA T, LAINE S, et al. Progressive growing of GANs for improved quality, stability, and variation[EB/OL]. [2024-12-28]. https://openreview.net/forum?id=Hk99zCeAb. https://openreview.net/forum?id=Hk99zCeAb
[21]
顾正华, 刘嘎琼, 邵长斌, . 深度检测方法中融合大小感受野机制的下采样算法[J]. 计算机科学与探索, 2024, 18(10): 2727-2737.
GU Z H, LIU G Q, SHAO C B, et al. Downsampling algorithm with fusion of different receptive field sizes in deep detection methods[J]. Journal of Frontiers of Technology, 2024, 18(10): 2727-2737 (in Chinese).
[22]
谢东升. 基于深度学习的车辆智能定损算法研究[D]. 天津: 天津大学, 2019.
XIE D S. Research on vehicle intelligent damage location algorithm based on deep learning[D]. Tianjin: Tianjin University, 2019 (in Chinese).
[23]
HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2018: 7132-7141.
[24]
JADERBERG M, SIMONYAN K, ZISSERMAN A. Spatial transformer networks[C]// The 29th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2015: 2017-2025.
[25]
WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]// The 15th European Conference on Computer Vision. Cham: Springer, 2018: 3-19.
[26]
GAO Y, ZENG Z, DU D, et al. SeerAttention: learning intrinsic sparse attention in your LLM[EB/OL]. [2024-12-28]. https://arxiv.org/abs/2410.13276. https://arxiv.org/abs/2410.13276
[27]
OUYANG D L, HE S, ZHANG G Z, et al. Efficient multi-scale attention module with cross-spatial learning[C]// ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing. New York: IEEE Press, 2023: 1-5.
[28]
ZHANG H, ZU K K, LU J, et al. EPSANet: an efficient pyramid squeeze attention block on convolutional neural network[C]// The 16th Asian Conference on Computer Vision. Cham: Springer, 2022: 541-557.
[29]
WANG Q L, WU B G, ZHU P F, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2020: 11531-11539.
[30]
CHEN L W, FU Y, WEI K X, et al. Instance segmentation in the dark[J]. International Journal of Computer Vision, 2023, 131(8): 2198-2218.
[31]
WANG F, JIANG M Q, QIAN C, et al. Residual attention network for image classification[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2017: 3156-3164.
[32]
WANG X K, LI W J, WU Z C. CarDD: a new dataset for vision-based car damage detection[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(7): 7202-7214.
[33]
ZHANG Z Z, LAN C L, ZENG W J, et al. Relation-aware global attention for person re-identification[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2020: 3183-3192.
[34]
HUANG H J, CHEN Z G, ZOU Y, et al. Channel prior convolutional attention for medical image segmentation[J]. Computers in Biology and Medicine, 2024, 178: 108784.
[35]
SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-cam: visual explanations from deep networks via gradient-based localization[C]// 2017 IEEE International Conference on Computer Vision. New York: IEEE Press, 2017: 618-626.
[36]
KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]// The 26th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2012: 1097-1105.
[37]
SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2015: 1-9.
[38]
HOWARD A G, ZHU M L, CHEN B, et al. MobileNets: efficient convolutional neural networks for mobile vision applications[EB/OL]. [2024-12-28]. https://arxiv.org/abs/1704.04861. https://arxiv.org/abs/1704.04861
[39]
ZHANG X Y, ZHOU X Y, LIN M X, et al. ShuffleNet: an extremely efficient convolutional neural network for mobile devices[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2018: 6848-6856.
[40]
HUANG G, LIU Z, VAN DER MAATEN L, et al. Densely connected convolutional networks[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2017: 2261-2269.
[41]
TAN M X, LE Q. EfficientNet: rethinking model scaling for convolutional neural networks[EB/OL]. [2024-12-28]. https://proceedings.mlr.press/v97/tan19a.html. https://proceedings.mlr.press/v97/tan19a.html
[42]
XU J, PAN Y, PAN X L, et al. RegNet: self-regulated network for image classification[J]. IEEE Transactions on Neural Networks and Learning Systems, 2023, 34(11): 9562-9567.
[43]
TAN M X, LE Q. EfficientNetV2:smaller models and faster training[EB/OL]. [2024-12-28]. https://proceedings.mlr.press/v139/tan21a. https://proceedings.mlr.press/v139/tan21a
[44]
CHEN J R, KAO S H, HE H, et al. Run, don't walk: chasing higher FLOPS for faster neural networks[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2023: 12021-12031.
[45]
DING X H, ZHANG X Y, HAN J G, et al. Scaling up your kernels to 31×31: revisiting large kernel design in CNNs[C]// 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2022: 11953-11965.
[46]
MA X, DAI X Y, BAI Y, et al. Rewrite the stars[C]// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2024: 5694-5703.
[47]
DOSOVITSKIY A, BEYER L, KOLESNIKOV A, et al. An image is worth 16x16 words:transformers for image recognition at scale[EB/OL]. [2024-12-28]. https://openreview.net/forum?id=YicbFdNTTy. https://openreview.net/forum?id=YicbFdNTTy
[48]
LIU Z, LIN Y T, CAO Y, et al. Swin transformer: hierarchical vision transformer using shifted windows[C]// 2021 IEEE/CVF International Conference on Computer Vision. New York: IEEE Press, 2021: 9992-10002.
[49]
MEHTA S, RASTEGARI M. MobileViT: light-weight, general-purpose, and mobile-friendly vision transformer[EB/OL]. [2024-12-28]. https://openreview.net/forum?id=vh-0sUt8HlG. https://openreview.net/forum?id=vh-0sUt8HlG
2026年第47卷第1期
PDF下载
5
1
引用本文
BibTeX
文章信息
doi: 10.11996/JG.j.2095-302X.2026010017
  • 接收时间:2025-02-28
  • 首发时间:2026-05-19
  • 出版时间:2026-02-28
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2025-02-28
  • 录用日期:2025-06-23
基金
National Natural Science Foundation of China(62373151)
国家自然科学基金(62373151)
Natural Science Foundation of Hebei Province(F2023502010)
河北省自然科学基金(F2023502010)
Joint Funds of the National Natural Science Foundation of China(U21A20486)
国家自然科学基金联合基金项目重点支持项目(U21A20486)
Project for Cultivating the Innovative Ability of Full-time Postgraduate Students Studying in Hebei Province(CXZZSS2025152)
河北省在读研究生创新能力培养资助项目(CXZZSS2025152)
作者信息
    华北电力大学自动化系, 河北 保定 071003

通讯作者:

王乾铭,E-mail:
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/txxb/CN/10.11996/JG.j.2095-302X.2026010017
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