Article(id=1322949834885443622, tenantId=1146029695717560320, journalId=1249024232475115590, issueId=1322949743743213946, articleNumber=null, orderNo=null, doi=10.11834/jig.250354, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1753718400000, receivedDateStr=2025-07-29, revisedDate=1764950400000, revisedDateStr=2025-12-06, acceptedDate=null, acceptedDateStr=null, onlineDate=1793345419856, onlineDateStr=2026-10-30, pubDate=1784131200000, pubDateStr=2026-07-16, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1793345419856, onlineIssueDateStr=2026-10-30, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1793345419856, creator=13701087609, updateTime=1793345419856, updator=13701087609, issue=Issue{id=1322949743743213946, tenantId=1146029695717560320, journalId=1249024232475115590, year='2026', volume='31', issue='7', pageStart='2279', pageEnd='2686', issueExtLink='null', onlineDate='null', pubDate='1784131200000', pubDateStr='2026-07-16', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1793345398125, creator='13701087609', updateTime=1793348592997, updator='13701087609', preIssue=null, nextIssue=null, articleTotal=null, ext={EN=IssueExt(id=1322963144217330568, tenantId=1146029695717560320, journalId=1249024232475115590, issueId=1322949743743213946, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1322963144217330569, tenantId=1146029695717560320, journalId=1249024232475115590, issueId=1322949743743213946, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null, downloadFileDto=null}, startPage=2614, endPage=2630, ext={EN=ArticleExt(id=1322949835116130343, articleId=1322949834885443622, tenantId=1146029695717560320, journalId=1249024232475115590, language=EN, title=Highway parking violation detection by integrating dual-attention and spatiotemporal features, columnId=1249044019595125147, journalTitle=Journal of Image and Graphics, columnName=Image Analysis and Recognition, runingTitle=null, highlight=null, articleAbstract=
Objective

Automatic detection of illegal parking on highways through surveillance video is a crucial task for modern traffic management systems because it directly affects traffic flow,public safety,and operational efficiency. However,detecting parking violations in highway environments presents substantial challenges because of complex traffic dynamics,varying lighting conditions,and broad surveillance coverage. These challenges include frequent vehicle occlusions in dense traffic that lead to identity confusion and incorrect tracking,substantial perspective distortion that causes slow-moving vehicles to be mistakenly identified as parked,and degraded detection performance under low-illumination conditions,where vehicle features become blurred and are prone to false detection because of environmental lighting effects.

Method

To address these problems,this study proposes a novel highway illegal parking detection algorithm that integrates dual-attention feature fusion and spatiotemporal behavior analysis. The approach introduces a dual-attention fusion module to enhance multiscale feature representations extracted from YOLOv8,which are used as appearance descriptors in a vehicle tracking framework on the basis of BotSORT-ReID. This framework directly leverages YOLOv8’s multiscale feature outputs and enhances them with attention mechanisms in spatial and channel dimensions. The spatial attention module emphasizes critical regions related to vehicle targets,suppressing background noise,whereas the channel attention mechanism selectively emphasizes discriminative feature channels relevant to object identity. These enhancements substantially improve the robustness and accuracy of vehicle tracking,especially in scenarios involving occlusion,background clutter,or varying viewing angles. The second component of the algorithm focuses on violation judgment through spatiotemporal feature accumulation. Specifically,it uses an adaptive interframe displacement estimation method on the basis of exponentially weighted moving averages to identify low-speed or stationary vehicles over a sequence of frames. A normalized displacement model is employed to correct perspective distortions across different regions of the image,ensuring that movement estimates are reliable even when vehicle scale varies with the distance from the camera. This method smooths short-term fluctuations in object position and helps avoid misclassification caused by detection box jitter or gradual deceleration. A secondary validation mechanism is proposed to address the complexity of nighttime detection,especially in environments with poor visibility or interference from artificial lights. This process involves analyzing the temporal characteristics of brightness changes to identify flashing hazard lights,which are a strong indicator of stationary vehicles in emergency situations. Time-series analysis is conducted on grayscale intensity values extracted from target bounding boxes across frames. Preprocessing steps,including Savitzky-Golay filtering,detrending,and temporal gradient computation,are applied to isolate periodic brightness variations. Subsequently,Welch’s method is used for power spectral density estimation,and dominant frequencies are extracted and compared with known hazard light flashing frequency ranges. Only when these frequencies match expected patterns is the object confirmed to be illegally parked,thereby reducing false positives from nonvehicle light sources or reflections.

Result

Given the absence of publicly available datasets for highway parking violation detection,three dedicated highway datasets are constructed for object detection,vehicle tracking,and parking violation recognition. The datasets consist of 8 532 labeled images for detection,34 600 frames for tracking,and 77 annotated video sequences (average over 1 min,covering 98 illegal parking vehicles) for violation recognition. Moreover,they cover diverse conditions,including daytime,nighttime,tunnels,and adverse weather (e.g.,rain and fog),thereby ensuring comprehensive evaluation across complex highway scenarios. Experimental results on the constructed datasets show that the proposed algorithm consistently outperforms several benchmark methods in terms of precision,recall,and F1 score. Notably,dual-attention feature extraction improves identity consistency in vehicle tracking,thereby reducing ID switches and increasing tracking accuracy. The adaptive displacement method demonstrates strong resistance to jitter and perspective artifacts,and the hazard light detection module substantially reduces nighttime false positives and improves robustness under difficult lighting. Ablation studies further confirm the contribution of each component. For instance,removing the dual-attention module results in reduced tracking quality and an increase in identity confusion. Without the adaptive displacement method,the system becomes highly vulnerable to misclassification because of minor motion or camera instability. Meanwhile,excluding the temporal signal-based hazard light detection leads to a rise in false alarms during nighttime evaluations. The full system achieves a balanced trade-off between high precision and high recall,demonstrating its effectiveness in real-world applications.

Conclusion

Overall,the proposed algorithm presents a unified,efficient,accurate solution for illegal parking detection on highways. By combining deep learning-based object detection,attention-guided feature fusion,motion pattern recognition,and signal-based validation,this algorithm addresses the key limitations of current methods and achieves robust performance across varying highway surveillance conditions. The integration of lightweight attention modules into the YOLOv8 backbone and the design of spatiotemporal behavioral analysis allow for real-time processing while maintaining high accuracy. The system is thus well-suited for deployment in intelligent transportation systems and smart city infrastructure where reliability and adaptability are paramount. The methodology also provides a scalable framework for extending similar detection strategies to other traffic violations or abnormal behaviors in complex road environments.

, authors=Dian Liu1, Na Yin1, Zening Zhang2, Hongyan Lyu2, Xubin Luo2, Haoxiang Wang1, *, authorsList=Dian Liu, Na Yin, Zening Zhang, Hongyan Lyu, Xubin Luo, Haoxiang Wang, authorCompany=null, correspAuthors=Haoxiang Wang, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, fund=null), CN=ArticleExt(id=1322949840359010376, articleId=1322949834885443622, tenantId=1146029695717560320, journalId=1249024232475115590, language=CN, title=融合双重注意力特征与时空特征的高速公路违停识别, columnId=1249044019838394781, journalTitle=中国图象图形学报, columnName=图像分析和识别, runingTitle=null, highlight=null, articleAbstract=
目的

为提升高速公路交通管理效率与道路安全,提出一种面向监控视频的违法停车自动识别方法,以解决车流密集场景中目标遮挡、摄像机透视畸变及夜间光照干扰导致的身份混淆与误判问题。

方法

引入了双重注意力特征融合结构,增强YOLOv8(you only look once)提取的多尺度特征的表达能力,并将生成的外观特征用于后续的轨迹匹配阶段,从而在保持高效跟踪性能的同时,有效缓解了目标身份混淆与切换的现象。此外,还设计了一种基于帧间位移自适应的判别机制来识别违停行为,有效降低了由视角畸变和边框抖动造成的误判和漏判风险。针对夜间检测的困难,进一步引入基于时间序列信号分析的危险报警闪光灯检测技术,以提升夜间判定的稳定性与准确性,从而增强整体算法的鲁棒性。

结果

鉴于当前尚无公开可用的高速公路违停检测数据集,结合实际交通管理需求,自主构建覆盖多场景的专用数据集,并在此基础上进行实验验证。结果表明,所提方法能够良好适应不同时间段及复杂条件下的高速公路场景,违停识别准确率达94.79%,召回率达92.86%,F1分数为0.94,显著优于对比方法。所提方法的图像处理速度为32帧/s,满足实时处理需求。

结论

提出的融合双重注意力特征与时空特征的方法,在高速公路违停识别上成效显著,该研究具备显著的开创性和实践指导意义。

, authors=刘电1, 尹娜1, 张泽宁2, 吕洪燕2, 罗旭彬2, 王昊翔1, *, authorsList=刘电, 尹娜, 张泽宁, 吕洪燕, 罗旭彬, 王昊翔, authorCompany=null, correspAuthors=王昊翔, authorNote=

刘电,男,博士研究生,主要研究方向为深度学习和目标检测与语义分割。E-mail:

尹娜,女,硕士研究生,主要研究方向为计算机视觉和目标检测。E-mail:

张泽宁,男,软件工程师,主要研究方向为计算机视觉、自然语言处理、生成式模型以及多模态学习。E-mail:

吕洪燕,男,正高级工程师,主要研究方向为智能交通、交通管控和深度学习。E-mail:

罗旭彬,男,软件工程师,主要研究方向为计算机视觉、自然语言处理和模式识别。E-mail:

王昊翔,通信作者,男,副教授,硕士生导师,主要研究方向为机器视觉与模式识别。E-mail:

, correspAuthorsNote=
王昊翔
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=KJYar+YbYfH5Ov1fnGlC6Q==, magXml=Im3QrTIQ/u/MajUurtqP7w==, pdfUrl=null, pdf=AdNSXHXdm7gd6joPLNkbgg==, pdfFileSize=7406929, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=NDnQ2M9hiJy13RNyzX93TQ==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=E2gKUGI2rfYcoj5TaLukyA==, mapNumber=null, fund=null)}, authors=[Author(id=1322949842305167440, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=csdianl@mail.scut.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949842393247826, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842305167440, language=EN, stringName=Dian Liu, firstName=Dian, middleName=null, lastName=Liu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949842485522515, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842305167440, language=CN, stringName=刘电, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 华南理工大学计算机科学与工程学院,广州 510006, bio={"content":"

刘电,男,博士研究生,主要研究方向为深度学习和目标检测与语义分割。E-mail:

"}, bioImg=null, bioContent=

刘电,男,博士研究生,主要研究方向为深度学习和目标检测与语义分割。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949840627445833, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=1, ext=[AuthorCompanyExt(id=1322949840635834442, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China), AuthorCompanyExt(id=1322949840648417355, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 华南理工大学计算机科学与工程学院,广州 510006)])]), Author(id=1322949842561019989, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=202220143482@mail.scut.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949842653294679, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842561019989, language=EN, stringName=Na Yin, firstName=Na, middleName=null, lastName=Yin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949842724597848, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842561019989, language=CN, stringName=尹娜, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 华南理工大学计算机科学与工程学院,广州 510006, bio={"content":"

尹娜,女,硕士研究生,主要研究方向为计算机视觉和目标检测。E-mail:

"}, bioImg=null, bioContent=

尹娜,女,硕士研究生,主要研究方向为计算机视觉和目标检测。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949840627445833, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=1, ext=[AuthorCompanyExt(id=1322949840635834442, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China), AuthorCompanyExt(id=1322949840648417355, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 华南理工大学计算机科学与工程学院,广州 510006)])]), Author(id=1322949842833649754, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=zznmizzou@gmail.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949842925924444, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842833649754, language=EN, stringName=Zening Zhang, firstName=Zening, middleName=null, lastName=Zhang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949843001421917, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842833649754, language=CN, stringName=张泽宁, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 广州国交润万交通信息有限公司,广州 510665, bio={"content":"

张泽宁,男,软件工程师,主要研究方向为计算机视觉、自然语言处理、生成式模型以及多模态学习。E-mail:

"}, bioImg=null, bioContent=

张泽宁,男,软件工程师,主要研究方向为计算机视觉、自然语言处理、生成式模型以及多模态学习。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949842191921228, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=2, ext=[AuthorCompanyExt(id=1322949842200309837, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China), AuthorCompanyExt(id=1322949842212892750, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 广州国交润万交通信息有限公司,广州 510665)])]), Author(id=1322949843261468767, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=lhy@runonetech.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949843345354849, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949843261468767, language=EN, stringName=Hongyan Lyu, firstName=Hongyan, middleName=null, lastName=Lyu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949843433435234, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949843261468767, language=CN, stringName=吕洪燕, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 广州国交润万交通信息有限公司,广州 510665, bio={"content":"

吕洪燕,男,正高级工程师,主要研究方向为智能交通、交通管控和深度学习。E-mail:

"}, bioImg=null, bioContent=

吕洪燕,男,正高级工程师,主要研究方向为智能交通、交通管控和深度学习。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949842191921228, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=2, ext=[AuthorCompanyExt(id=1322949842200309837, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China), AuthorCompanyExt(id=1322949842212892750, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 广州国交润万交通信息有限公司,广州 510665)])]), Author(id=1322949843508932708, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=515664321@qq.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949843739619430, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949843508932708, language=EN, stringName=Xubin Luo, firstName=Xubin, middleName=null, lastName=Luo, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949843815116903, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949843508932708, language=CN, stringName=罗旭彬, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 广州国交润万交通信息有限公司,广州 510665, bio={"content":"

罗旭彬,男,软件工程师,主要研究方向为计算机视觉、自然语言处理和模式识别。E-mail:

"}, bioImg=null, bioContent=

罗旭彬,男,软件工程师,主要研究方向为计算机视觉、自然语言处理和模式识别。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949842191921228, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=2, ext=[AuthorCompanyExt(id=1322949842200309837, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China), AuthorCompanyExt(id=1322949842212892750, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 广州国交润万交通信息有限公司,广州 510665)])]), Author(id=1322949844087746665, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=5, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=hxwang@scut.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1322949844171632747, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949844087746665, language=EN, stringName=Haoxiang Wang, firstName=Haoxiang, middleName=null, lastName=Wang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, *, address=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949844263907436, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949844087746665, language=CN, stringName=王昊翔, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, *, address=1 华南理工大学计算机科学与工程学院,广州 510006, bio={"content":"

王昊翔,通信作者,男,副教授,硕士生导师,主要研究方向为机器视觉与模式识别。E-mail:

"}, bioImg=null, bioContent=

王昊翔,通信作者,男,副教授,硕士生导师,主要研究方向为机器视觉与模式识别。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949840627445833, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=1, ext=[AuthorCompanyExt(id=1322949840635834442, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China), AuthorCompanyExt(id=1322949840648417355, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 华南理工大学计算机科学与工程学院,广州 510006)])])], keywords=[Keyword(id=1322949844402319469, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, orderNo=1, keyword=highway), Keyword(id=1322949844515565678, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, orderNo=2, keyword=spatiotemporal feature), Keyword(id=1322949844595257455, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, orderNo=3, keyword=attention feature), Keyword(id=1322949844687532144, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, orderNo=4, keyword=parking violation detection), Keyword(id=1322949844825944177, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, orderNo=5, keyword=dual attention), Keyword(id=1322949844960161906, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, orderNo=1, keyword=高速公路), Keyword(id=1322949846633689203, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, orderNo=2, keyword=时空特征), Keyword(id=1322949846713380980, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, orderNo=3, keyword=注意力特征), Keyword(id=1322949846780489845, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, orderNo=4, keyword=违停识别), Keyword(id=1322949846855987318, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, orderNo=5, keyword=双重注意力)], refs=[Reference(id=1322949852799316139, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=null, pageEnd=null, url=http://arxiv.org/pdf/2206.14651v1.pdf, language=null, rfNumber=null, rfOrder=0, authorNames=Aharon N, Orfaig R, Bobrovsky B Z, journalName=null, refType=null, unstructuredReference=Aharon N,Orfaig R and Bobrovsky B Z. 2022. BoT-SORT: robust associations multi-pedestrian tracking [EB/OL]. [2025-07-15]., articleTitle=BoT-SORT: robust associations multi-pedestrian tracking, refAbstract=null), Reference(id=1322949852879007916, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2024, volume=50, issue=4, pageStart=11, pageEnd=19, url=null, language=null, rfNumber=null, rfOrder=1, authorNames=Chen W, Wang X L, Zhang Y W, An G C, Jiang B, journalName=Computer Engineering, refType=null, unstructuredReference=Chen W,Wang X L,Zhang Y W,An G C and Jiang B. 2024. Vehicle illegal parking detection in highway service areas based on improved YOLOv8. Computer Engineering,50(4): 11-19, articleTitle=Vehicle illegal parking detection in highway service areas based on improved YOLOv8, refAbstract=null), Reference(id=1322949852958699693, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2024, volume=50, issue=4, pageStart=11, pageEnd=19, url=null, language=null, rfNumber=null, rfOrder=2, authorNames=陈伟, 王晓龙, 张晏玮, 安国成, 江波, journalName=计算机工程, refType=null, unstructuredReference=陈伟,王晓龙,张晏玮,安国成,江波. 2024. 基于改进YOLOv8的高速公路服务区车辆违停检测. 计算机工程,50(4): 11-19[DOI: 10.19678/j.issn.1000-3428.0068901], articleTitle=基于改进YOLOv8的高速公路服务区车辆违停检测, refAbstract=null), Reference(id=1322949853017419950, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2023, volume=25, issue=null, pageStart=8725, pageEnd=8737, url=null, language=null, rfNumber=null, rfOrder=3, authorNames=Du Y H, Zhao Z C, Song Y, Zhao Y Y, Su F, Gong T, journalName=IEEE Transactions on Multimedia, refType=null, unstructuredReference=Du Y H,Zhao Z C,Song Y,Zhao Y Y,Su F,Gong T,et al. 2023. StrongSORT: make DeepSORT great again. IEEE Transactions on Multimedia,25: 8725-8737 [DOI: 10.1109/TMM.2023.3240881], articleTitle=StrongSORT: make DeepSORT great again, refAbstract=null), Reference(id=1322949853092917423, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=169, pageEnd=174, url=null, language=null, rfNumber=null, rfOrder=4, authorNames=Hamzah R A, Setianingsih C, Nugrahaeni R A, Hanafia S R, Fuadi F, journalName=null, refType=null, unstructuredReference=Hamzah R A,Setianingsih C,Nugrahaeni R A,Hanafia S R and Fuadi F. 2022. Parking violation detection on the roadside of toll roads with intelligent transportation system using faster R-CNN algorithm//Proceedings of the 6th International Conference on Informatics and Computational Sciences. Semarang,Indonesia: IEEE: 169-174 [DOI: 10.1109/ICICoS56336.2022.9930590], articleTitle=Parking violation detection on the roadside of toll roads with intelligent transportation system using faster R-CNN algorithm, refAbstract=null), Reference(id=1322949853168414896, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2012, volume=10, issue=5, pageStart=972, pageEnd=981, url=null, language=null, rfNumber=null, rfOrder=5, authorNames=Hassan W, Birch P, Young R, Chatwin C, journalName=International Journal of Control,Automation and Systems, refType=null, unstructuredReference=Hassan W,Birch P,Young R and Chatwin C. 2012. Real-time occlusion tolerant detection of illegally parked vehicles. International Journal of Control,Automation and Systems,10(5): 972-981 [DOI: 10.1007/s12555-012-0514-2], articleTitle=Real-time occlusion tolerant detection of illegally parked vehicles, refAbstract=null), Reference(id=1322949853252300977, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2020, volume=42, issue=8, pageStart=2011, pageEnd=2023, url=null, language=null, rfNumber=null, rfOrder=6, authorNames=Hu J, Shen L, Albanie S, Sun G, Wu E H, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=Hu J,Shen L,Albanie S,Sun G and Wu E H. 2020. Squeeze-and-excitation networks. IEEE Transactions on Pattern Analysis and Machine Intelligence,42(8): 2011-2023 [DOI: 10.1109/TPAMI.2019.2913372], articleTitle=Squeeze-and-excitation networks, refAbstract=null), Reference(id=1322949853319409842, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=448, pageEnd=456, url=null, language=null, rfNumber=null, rfOrder=7, authorNames=Ioffe S, Szegedy C, journalName=null, refType=null, unstructuredReference=Ioffe S and Szegedy C. 2015. Batch normalization: accelerating deep network training by reducing internal covariate shift//Proceedings of the 32nd International Conference on Machine Learning. Lille,France: JMLR.org: 448-456, articleTitle=Batch normalization: accelerating deep network training by reducing internal covariate shift, refAbstract=null), Reference(id=1322949853386518707, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=538, pageEnd=542, url=null, language=null, rfNumber=null, rfOrder=8, authorNames=Ketcham M, Getkhaw E, Piyaneeranart M, Ganokratanaa T, Yimyam W, journalName=null, refType=null, unstructuredReference=Ketcham M,Getkhaw E,Piyaneeranart M,Ganokratanaa T and Yimyam W. 2019. Recognizing the illegal parking patterns of cars on the road in front of the bus stop using the support vector machine//Proceedings of the 15th International Conference on Signal-Image Technology and Internet-Based Systems. Sorrento,Italy: IEEE: 538-542 [DOI: 10.1109/SITIS.2019.00091], articleTitle=Recognizing the illegal parking patterns of cars on the road in front of the bus stop using the support vector machine, refAbstract=null), Reference(id=1322949853453627572, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=1865, pageEnd=1870, url=null, language=null, rfNumber=null, rfOrder=9, authorNames=Kuo L C, Lin H Y, journalName=Proceedings of 2024 IEEE Intelligent Vehicles Symposium, refType=null, unstructuredReference=Kuo L C and Lin H Y. 2024. Illegal parking detection based on multi-task driving perception//Proceedings of 2024 IEEE Intelligent Vehicles Symposium. Jeju Island,Korea (South): IEEE: 1865-1870 [DOI: 10.1109/iv55156.2024.10588798], articleTitle=Illegal parking detection based on multi-task driving perception, refAbstract=null), Reference(id=1322949853524930741, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2022, volume=39, issue=6, pageStart=1916, pageEnd=1920, url=null, language=null, rfNumber=null, rfOrder=10, authorNames=Liang R L, Wang R, Guo Y, journalName=Application Research of Computers, refType=null, unstructuredReference=Liang R L,Wang R and Guo Y. 2022. Highway abnormal parking detection based on video spatio-temporal relationship. Application Research of Computers,39(6): 1916-1920, articleTitle=Highway abnormal parking detection based on video spatio-temporal relationship, refAbstract=null), Reference(id=1322949853592039606, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2022, volume=39, issue=6, pageStart=1916, pageEnd=1920, url=null, language=null, rfNumber=null, rfOrder=11, authorNames=梁睿琳, 王锐, 郭迎, journalName=计算机应用研究, refType=null, unstructuredReference=梁睿琳,王锐,郭迎. 2022. 基于视频时空关系的高速公路异常停车检测. 计算机应用研究,39(6): 1916-1920[DOI: 10.19734/j.issn.1001-3695.2021.09.0617], articleTitle=基于视频时空关系的高速公路异常停车检测, refAbstract=null), Reference(id=1322949853654954167, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2021, volume=129, issue=2, pageStart=548, pageEnd=578, url=null, language=null, rfNumber=null, rfOrder=12, authorNames=Luiten J, Ošep A, Dendorfer P, Torr P, Geiger A, Leal-Taixé L, journalName=International Journal of Computer Vision, refType=null, unstructuredReference=Luiten J,Ošep A,Dendorfer P,Torr P,Geiger A,Leal-Taixé L,et al. 2021. HOTA: a higher order metric for evaluating multi-object tracking. International Journal of Computer Vision,129(2): 548-578 [DOI: 10.1007/s11263-020-01375-2], articleTitle=HOTA: a higher order metric for evaluating multi-object tracking, refAbstract=null), Reference(id=1322949853722063032, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=3025, pageEnd=3029, url=null, language=null, rfNumber=null, rfOrder=13, authorNames=Maggiolino G, Ahmad A, Cao J K, Kitani K, journalName=null, refType=null, unstructuredReference=Maggiolino G,Ahmad A,Cao J K and Kitani K. 2023. Deep OC-SORT: multi-pedestrian tracking by adaptive re-identification//Proceedings of 2023 IEEE International Conference on Image Processing. Kuala Lumpur,Malaysia: IEEE: 3025-3029 [DOI: 10.1109/ICIP49359.2023.10222576], articleTitle=Deep OC-SORT: multi-pedestrian tracking by adaptive re-identification, refAbstract=null), Reference(id=1322949853801754809, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=227, pageEnd=232, url=null, language=null, rfNumber=null, rfOrder=14, authorNames=Matsuda A, Matsui T, Matsuda Y, Suwa H, Yasumoto K, journalName=null, refType=null, unstructuredReference=Matsuda A,Matsui T,Matsuda Y,Suwa H and Yasumoto K. 2021. A system for real-time on-street parking detection and visualization on an edge device//Proceedings of 2021 IEEE International Conference on Pervasive Computing and Communications Workshops and Other Affiliated Events. Kassel,Germany: IEEE: 227-232 [DOI: 10.1109/PerComWorkshops51409.2021.9431076], articleTitle=A system for real-time on-street parking detection and visualization on an edge device, refAbstract=null), Reference(id=1322949853868863674, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2025, volume=13, issue=null, pageStart=7883, pageEnd=7901, url=null, language=null, rfNumber=null, rfOrder=15, authorNames=Park H, Kim K, Jeong I, Jung J, Cho J, journalName=IEEE Access, refType=null, unstructuredReference=Park H,Kim K,Jeong I,Jung J and Cho J. 2025. Special vehicle classification algorithm-based system for dedicated parking zone violation detection in South Korea. IEEE Access,13: 7883-7901 [DOI: 10.1109/ACCESS.2025.3526862], articleTitle=Special vehicle classification algorithm-based system for dedicated parking zone violation detection in South Korea, refAbstract=null), Reference(id=1322949853931778235, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2022, volume=23, issue=11, pageStart=20572, pageEnd=20587, url=null, language=null, rfNumber=null, rfOrder=16, authorNames=Peng X G, Song R Z H, Cao Q, Li Y, Cui D S, Jia X F, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=Peng X G,Song R Z H,Cao Q,Li Y,Cui D S,Jia X F,et al. 2022. Real-time illegal parking detection algorithm in urban environments. IEEE Transactions on Intelligent Transportation Systems,23(11): 20572-20587 [DOI: 10.1109/TITS.2022.3180225], articleTitle=Real-time illegal parking detection algorithm in urban environments, refAbstract=null), Reference(id=1322949855613694140, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2007, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=17, authorNames=Press W H, Teukolsky S A, Vetterling W T, Flannery B P, journalName=null, refType=null, unstructuredReference=Press W H,Teukolsky S A,Vetterling W T and Flannery B P. 2007. Numerical Recipes: The Art of Scientific Computing. 3rd ed. Cambridge,UK: Cambridge University Press [DOI: 10.1007/BF01321860], articleTitle=Numerical Recipes: The Art of Scientific Computing. 3rd ed. Cambridge,UK: Cambridge University Press, refAbstract=null), Reference(id=1322949857555656893, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2017, volume=39, issue=6, pageStart=1137, pageEnd=1149, url=null, language=null, rfNumber=null, rfOrder=18, authorNames=Ren S Q, He K M, Girshick R, Sun J, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=Ren S Q,He K M,Girshick R and Sun J. 2017. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence,39(6): 1137-1149 [DOI: 10.1109/TPAMI.2016.2577031], articleTitle=Faster R-CNN: towards real-time object detection with region proposal networks, refAbstract=null), Reference(id=1322949857618571454, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=http://arxiv.org/pdf/2408.13003v1.pdf, language=null, rfNumber=null, rfOrder=19, authorNames=Stanojević V and Todorović B, journalName=null, refType=null, unstructuredReference=Stanojević V and Todorović B. 2024. BoostTrack++: using tracklet information to detect more objects in multiple object tracking [EB/OL]. [2025-07-15]., articleTitle=BoostTrack++: using tracklet information to detect more objects in multiple object tracking, refAbstract=null), Reference(id=1322949857694068927, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=772, pageEnd=775, url=null, language=null, rfNumber=null, rfOrder=20, authorNames=Wahyono, Filonenko A and Jo K H, journalName=null, refType=null, unstructuredReference=Wahyono,Filonenko A and Jo K H. 2016. Detecting illegally parked vehicle based on cumulative dual foreground difference//Proceedings of the 14th IEEE International Conference on Industrial Informatics. Poitiers,France: IEEE: 772-775 [DOI: 10.1109/INDIN.2016.7819263], articleTitle=Detecting illegally parked vehicle based on cumulative dual foreground difference, refAbstract=null), Reference(id=1322949857924755648, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2025, volume=269, issue=null, pageStart=null, pageEnd=126440, url=null, language=null, rfNumber=null, rfOrder=21, authorNames=Wang H, Liu J, Zhao J, Zhang J, Zhao D, journalName=Expert Systems with Applications, refType=null, unstructuredReference=Wang H,Liu J,Zhao J,Zhang J and Zhao D. 2025. Precision and speed: LSOD-YOLO for lightweight small object detection. Expert Systems with Applications,269: #126440 [DOI: 10.1016/j.eswa.2025.126440], articleTitle=Precision and speed: LSOD-YOLO for lightweight small object detection, refAbstract=null), Reference(id=1322949858004447425, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2024, volume=29, issue=5, pageStart=1364, pageEnd=1380, url=null, language=null, rfNumber=null, rfOrder=22, authorNames=Wang Q, Xue X Y, Min W D, Wang S, Gai D, Han Q, journalName=Journal of Image and Graphics, refType=null, unstructuredReference=Wang Q,Xue X Y,Min W D,Wang S,Gai D and Han Q. 2024. Cross-domain joint learning and shared subspace metric for vehicle re-identification. Journal of Image and Graphics,29(5): 1364-1380, articleTitle=Cross-domain joint learning and shared subspace metric for vehicle re-identification, refAbstract=null), Reference(id=1322949858092527810, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2024, volume=29, issue=5, pageStart=1364, pageEnd=1380, url=null, language=null, rfNumber=null, rfOrder=23, authorNames=汪琦, 雪心远, 闵卫东, 汪晟, 盖迪, 韩清, journalName=中国图象图形学报, refType=null, unstructuredReference=汪琦,雪心远,闵卫东,汪晟,盖迪,韩清. 2024. 跨域联合学习与共享子空间度量的车辆重识别. 中国图象图形学报,29(5): 1364-1380 [DOI: 10.11834/jig.230405], articleTitle=跨域联合学习与共享子空间度量的车辆重识别, refAbstract=null), Reference(id=1322949858344186051, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=1967, volume=15, issue=2, pageStart=70, pageEnd=73, url=null, language=null, rfNumber=null, rfOrder=24, authorNames=Welch P, journalName=IEEE Transactions on Audio and Electroacoustics, refType=null, unstructuredReference=Welch P. 1967. The use of fast Fourier transform for the estimation of power spectra: a method based on time averaging over short,modified periodograms. IEEE Transactions on Audio and Electroacoustics,15(2): 70-73 [DOI: 10.1109/TAU.1967.1161901], articleTitle=The use of fast Fourier transform for the estimation of power spectra: a method based on time averaging over short,modified periodograms, refAbstract=null), Reference(id=1322949858428072132, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=3, pageEnd=19, url=null, language=null, rfNumber=null, rfOrder=25, authorNames=Woo S, Park J, Lee J Y, Kweon I S, journalName=null, refType=null, unstructuredReference=Woo S,Park J,Lee J Y and Kweon I S. 2018. CBAM: convolutional block attention module//Proceedings of the 15th European Conference on Computer Vision. Munich,Germany: Springer: 3-19 [DOI: 10.1007/978-3-030-01234-2_1], articleTitle=CBAM: convolutional block attention module, refAbstract=null), Reference(id=1322949860080627910, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=4082, pageEnd=4091, url=null, language=null, rfNumber=null, rfOrder=26, authorNames=Xiao C C, Cao Q, Luo Z G, Lan L, journalName=null, refType=null, unstructuredReference=Xiao C C,Cao Q,Luo Z G and Lan L. 2024. MambaTrack: a simple baseline for multiple object tracking with state space model//Proceedings of the 32nd ACM International Conference on Multimedia. Melbourne,Australia: ACM: 4082-4091 [DOI: 10.1145/3664647.3680944], articleTitle=MambaTrack: a simple baseline for multiple object tracking with state space model, refAbstract=null), Reference(id=1322949860164513991, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2017, volume=null, issue=null, pageStart=23, pageEnd=27, url=null, language=null, rfNumber=null, rfOrder=27, authorNames=Xie X M, Wang C Y, Chen S, Shi G M, Zhao Z F, journalName=null, refType=null, unstructuredReference=Xie X M,Wang C Y,Chen S,Shi G M and Zhao Z F. 2017. Real-time illegal parking detection system based on deep learning//Proceedings of 2017 International Conference on Deep Learning Technologies. Chengdu,China: ACM: 23-27 [DOI: 10.1145/3094243.3094261], articleTitle=Real-time illegal parking detection system based on deep learning, refAbstract=null), Reference(id=1322949860244205768, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2025, volume=30, issue=1, pageStart=212, pageEnd=224, url=null, language=null, rfNumber=null, rfOrder=28, authorNames=Xu H, Dong S H, Zhang J W, Zheng Y H, journalName=Journal of Image and Graphics, refType=null, unstructuredReference=Xu H,Dong S H,Zhang J W and Zheng Y H. 2025. Context-aware attention fused Transformer tracking. Journal of Image and Graphics,30(1): 212-224, articleTitle=Context-aware attention fused Transformer tracking, refAbstract=null), Reference(id=1322949860319703241, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2025, volume=30, issue=1, pageStart=212, pageEnd=224, url=null, language=null, rfNumber=null, rfOrder=29, authorNames=徐晗, 董仕豪, 张家伟, 郑钰辉, journalName=中国图象图形学报, refType=null, unstructuredReference=徐晗,董仕豪,张家伟,郑钰辉. 2025. 融合上下文感知注意力的Transformer目标跟踪方法. 中国图象图形学报,30(1): 212-224 [DOI: 10.11834/jig.240084], articleTitle=融合上下文感知注意力的Transformer目标跟踪方法, refAbstract=null), Reference(id=1322949860386812106, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2023, volume=28, issue=7, pageStart=2054, pageEnd=2067, url=null, language=null, rfNumber=null, rfOrder=30, authorNames=Yu Y, Chen W X, Chen F X, journalName=Journal of Image and Graphics, refType=null, unstructuredReference=Yu Y,Chen W X and Chen F X. 2023. RIC-NVNet: night-time vehicle enhancement network for vehicle model recognition. Journal of Image and Graphics,28(7): 2054-2067, articleTitle=RIC-NVNet: night-time vehicle enhancement network for vehicle model recognition, refAbstract=null), Reference(id=1322949860458115275, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2023, volume=28, issue=7, pageStart=2054, pageEnd=2067, url=null, language=null, rfNumber=null, rfOrder=31, authorNames=余烨, 陈维笑, 陈凤欣, journalName=中国图象图形学报, refType=null, unstructuredReference=余烨,陈维笑,陈凤欣. 2023. 面向车型识别的夜间车辆图像增强网络RIC-NVNet. 中国图象图形学报,28(7): 2054-2067[DOI: 10.11834/jig.220122], articleTitle=面向车型识别的夜间车辆图像增强网络RIC-NVNet, refAbstract=null), Reference(id=1322949860554584268, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=659, pageEnd=675, url=null, language=null, rfNumber=null, rfOrder=32, authorNames=Zeng F G, Dong B, Zhang Y, Wang T C, Zhang X Y, Wei Y C, journalName=null, refType=null, unstructuredReference=Zeng F G,Dong B,Zhang Y,Wang T C,Zhang X Y and Wei Y C. 2022. MOTR: end-to-end multiple-object tracking with Transformer//Proceedings of the 17th European Conference on Computer Vision. Tel Aviv,Israel: Springer: 659-675 [DOI: 10.1007/978-3-031-19812-0_38], articleTitle=MOTR: end-to-end multiple-object tracking with Transformer, refAbstract=null), Reference(id=1322949860630081741, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=1, pageEnd=21, url=null, language=null, rfNumber=null, rfOrder=33, authorNames=Zhang Y F, Sun P Z, Jiang Y, Yu D D, Weng F C, Yuan Z H, journalName=null, refType=null, unstructuredReference=Zhang Y F,Sun P Z,Jiang Y,Yu D D,Weng F C,Yuan Z H,et al. 2022. ByteTrack: multi-object tracking by associating every detection box//Proceedings of the 17th European Conference on Computer Vision. Tel Aviv,Israel: Springer: 1-21 [DOI: 10.1007/978-3-031-20047-2_1], articleTitle=ByteTrack: multi-object tracking by associating every detection box, refAbstract=null)], funds=[Fund(id=1322949852635738281, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, awardId=null, language=EN, fundingSource=Guangdong Provincial Science and Technology Project(2023A1111120018), fundOrder=null, country=null), Fund(id=1322949852702847146, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, awardId=null, language=CN, fundingSource=广东省省级科技计划项目(2023A1111120018), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1322949840627445833, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=1, ext=[AuthorCompanyExt(id=1322949840635834442, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China), AuthorCompanyExt(id=1322949840648417355, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 华南理工大学计算机科学与工程学院,广州 510006)]), AuthorCompany(id=1322949842191921228, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=2, ext=[AuthorCompanyExt(id=1322949842200309837, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China), AuthorCompanyExt(id=1322949842212892750, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 广州国交润万交通信息有限公司,广州 510665)])], figs=[ArticleFig(id=1322949847011176567, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.1, caption=Schematic diagram of vehicle illegal parking detection algorithm, figureFileSmall=9RasrNCeGAhGVMHgdFHeOg==, figureFileBig=NDnQ2M9hiJy13RNyzX93TQ==, tableContent=null), ArticleFig(id=1322949847086674040, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图1, caption=车辆违停识别算法示意图, figureFileSmall=9RasrNCeGAhGVMHgdFHeOg==, figureFileBig=NDnQ2M9hiJy13RNyzX93TQ==, tableContent=null), ArticleFig(id=1322949847296389241, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.2, caption=Structure of the dual attention feature fusion module, figureFileSmall=kfHNZX+SHu+nfNvOPyTwSg==, figureFileBig=xohuxePuNbFI+vEl+pK8VA==, tableContent=null), ArticleFig(id=1322949847371886714, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图2, caption=双重注意力特征融合模块结构, figureFileSmall=kfHNZX+SHu+nfNvOPyTwSg==, figureFileBig=xohuxePuNbFI+vEl+pK8VA==, tableContent=null), ArticleFig(id=1322949847434801275, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.3, caption=SEBlocks architecture diagram, figureFileSmall=C6g92UzNjjZxB7s5TKu8Ow==, figureFileBig=byIS9kfs0zxAK4sSCvIVbQ==, tableContent=null), ArticleFig(id=1322949847510298748, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图3, caption=SEBlocks结构图, figureFileSmall=C6g92UzNjjZxB7s5TKu8Ow==, figureFileBig=byIS9kfs0zxAK4sSCvIVbQ==, tableContent=null), ArticleFig(id=1322949847602573437, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.4, caption=Flowchart of the illegal parking detection method based on spatiotemporal features, figureFileSmall=ORBxo2ig798zRmB8Kh4yuQ==, figureFileBig=rFwjBFXFSQgqriGGNFIM6A==, tableContent=null), ArticleFig(id=1322949847728402558, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图4, caption=基于时空特征的违停判别方法流程图, figureFileSmall=ORBxo2ig798zRmB8Kh4yuQ==, figureFileBig=rFwjBFXFSQgqriGGNFIM6A==, tableContent=null), ArticleFig(id=1322949847803900031, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.5, caption=Distribution diagram of inter-frame displacement data for vehicle targets, figureFileSmall=v8aetb91bsC8dDbCv5y14g==, figureFileBig=9tQCjymjndLDxJ4hjtbMaw==, tableContent=null), ArticleFig(id=1322949847875203200, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图5, caption=车辆目标的帧间位移数据分布图, figureFileSmall=v8aetb91bsC8dDbCv5y14g==, figureFileBig=9tQCjymjndLDxJ4hjtbMaw==, tableContent=null), ArticleFig(id=1322949847946506369, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.6, caption=Schematic diagram of hazard warning flashing light temporal signal analysis, figureFileSmall=EwhZ72cEbxz/8N3k3OhopQ==, figureFileBig=ZQjWYgFRNgJklyDV58mwrg==, tableContent=null), ArticleFig(id=1322949848022003842, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图6, caption=危险报警闪光灯时序信号分析流程示意图, figureFileSmall=EwhZ72cEbxz/8N3k3OhopQ==, figureFileBig=ZQjWYgFRNgJklyDV58mwrg==, tableContent=null), ArticleFig(id=1322949848097501315, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.7, caption=Brightness curve preprocessing flow, figureFileSmall=ZEaK5j+B35CjQVvGF6RAOQ==, figureFileBig=9MpWh2Y4gPhAmwUX0ikNXg==, tableContent=null), ArticleFig(id=1322949848168804484, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图7, caption=亮度曲线预处理流程, figureFileSmall=ZEaK5j+B35CjQVvGF6RAOQ==, figureFileBig=9MpWh2Y4gPhAmwUX0ikNXg==, tableContent=null), ArticleFig(id=1322949848240107653, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.8, caption=Schematic diagram of curve spectrum analysis, figureFileSmall=4AD4/W8RCbOqnmTMY4ocGA==, figureFileBig=1RfaHq/7OnfzVm+g+4mQIA==, tableContent=null), ArticleFig(id=1322949848332382342, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图8, caption=曲线频谱分析示意图, figureFileSmall=4AD4/W8RCbOqnmTMY4ocGA==, figureFileBig=1RfaHq/7OnfzVm+g+4mQIA==, tableContent=null), ArticleFig(id=1322949848433045639, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.9, caption=Low-light suspected vehicle images, figureFileSmall=HOysDj8oA9cUDVLh8coK4w==, figureFileBig=YUDdxjZVOC/oFzgY+Z8Quw==, tableContent=null), ArticleFig(id=1322949848521126024, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图9, caption=暗光疑似车辆图像, figureFileSmall=HOysDj8oA9cUDVLh8coK4w==, figureFileBig=YUDdxjZVOC/oFzgY+Z8Quw==, tableContent=null), ArticleFig(id=1322949848592429193, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.10, caption=Average brightness temporal graph of suspected targets, figureFileSmall=7k++hgmeqwuWtBNXD6E3lg==, figureFileBig=Iap2x8H14V7HrRgMeTaVSg==, tableContent=null), ArticleFig(id=1322949848667926666, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图10, caption=疑似目标时序平均亮度图, figureFileSmall=7k++hgmeqwuWtBNXD6E3lg==, figureFileBig=Iap2x8H14V7HrRgMeTaVSg==, tableContent=null), ArticleFig(id=1322949848756007051, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.11, caption=Example images of four categories, figureFileSmall=q3efZRil2i7y3+G+fGPuVg==, figureFileBig=VqCksdhjoFe1cXz0nKxChQ==, tableContent=null), ArticleFig(id=1322949848831504524, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图11, caption=4类目标的示例图像, figureFileSmall=q3efZRil2i7y3+G+fGPuVg==, figureFileBig=VqCksdhjoFe1cXz0nKxChQ==, tableContent=null), ArticleFig(id=1322949848915390605, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.12, caption=Display of images from different scenarios in the vehicle illegal parking detection dataset, figureFileSmall=RfV1w90YMYHSaciWGIkcog==, figureFileBig=W2xPawjmw+vB0iA++sqJBA==, tableContent=null), ArticleFig(id=1322949848978305166, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图12, caption=车辆违停识别数据集不同场景图像展示, figureFileSmall=RfV1w90YMYHSaciWGIkcog==, figureFileBig=W2xPawjmw+vB0iA++sqJBA==, tableContent=null), ArticleFig(id=1322949849053802639, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.13, caption=Examples of partial vehicle illegal parking detection, figureFileSmall=zHIf17T8Hh4uAihpeG2MJQ==, figureFileBig=yprnh2kU41rIj/6i1OF4PA==, tableContent=null), ArticleFig(id=1322949849125105808, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图13, caption=车辆部分违停识别例子, figureFileSmall=zHIf17T8Hh4uAihpeG2MJQ==, figureFileBig=yprnh2kU41rIj/6i1OF4PA==, tableContent=null), ArticleFig(id=1322949849200603281, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.14, caption=Example diagram of experimental results for the effectiveness of the appearance feature extraction module, figureFileSmall=OFrULvLUx8Dgf6CTyEDA0g==, figureFileBig=eJ9U8GBOQljk92Yua6e3MA==, tableContent=null), ArticleFig(id=1322949849292877970, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图14, caption=表观特征提取模块有效性的实验结果示例图, figureFileSmall=OFrULvLUx8Dgf6CTyEDA0g==, figureFileBig=eJ9U8GBOQljk92Yua6e3MA==, tableContent=null), ArticleFig(id=1322949849372569747, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.15, caption=Example diagram of experimental results for adaptive inter-frame displacement, figureFileSmall=ACTL7lfBTR7umVR0sD30Gg==, figureFileBig=lZyoZe1+9wgBvLH21jMkYg==, tableContent=null), ArticleFig(id=1322949849439678612, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图15, caption=自适应帧间位移的实验结果示例图, figureFileSmall=ACTL7lfBTR7umVR0sD30Gg==, figureFileBig=lZyoZe1+9wgBvLH21jMkYg==, tableContent=null), ArticleFig(id=1322949851113205909, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Fig.16, caption=Example diagram of experimental results for secondary verification based on temporal signal analysis, figureFileSmall=LpqZ8EwM1Tde65qRo8dHTQ==, figureFileBig=L5HOyFYNnWMcIvh2EjN3kA==, tableContent=null), ArticleFig(id=1322949851205480598, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=图16, caption=基于时序信号分析的次级验证的实验结果示例图, figureFileSmall=LpqZ8EwM1Tde65qRo8dHTQ==, figureFileBig=L5HOyFYNnWMcIvh2EjN3kA==, tableContent=null), ArticleFig(id=1322949851285172375, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 1, caption=Statistical analysis of image data in highway vehicle detection dataset, figureFileSmall=null, figureFileBig=null, tableContent=
数据集

图像

数量

小型

车辆

大型

货车

大巴

车辆

暗光疑

似车辆

合计8 53237 0372 3011 6705 265
训练集5 99026 3081 6381 2643 610
测试集2 54210 7296634061 655
), ArticleFig(id=1322949851352281240, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表1, caption=

高速公路车辆检测数据集图像数据统计

, figureFileSmall=null, figureFileBig=null, tableContent=
数据集

图像

数量

小型

车辆

大型

货车

大巴

车辆

暗光疑

似车辆

合计8 53237 0372 3011 6705 265
训练集5 99026 3081 6381 2643 610
测试集2 54210 7296634061 655
), ArticleFig(id=1322949851436167321, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 2, caption=Statistical analysis of data in highway vehicle tracking dataset, figureFileSmall=null, figureFileBig=null, tableContent=
数据集片段数量图像总数轨迹总数
合计3634 6003 280
训练集2523 6842 143
测试集1110 9161 137
), ArticleFig(id=1322949851511664794, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表2, caption=

高速公路车辆跟踪数据集数据统计

, figureFileSmall=null, figureFileBig=null, tableContent=
数据集片段数量图像总数轨迹总数
合计3634 6003 280
训练集2523 6842 143
测试集1110 9161 137
), ArticleFig(id=1322949851587162267, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 3, caption=Statistical analysis of the number of scene segments in the vehicle illegal parking detection dataset, figureFileSmall=null, figureFileBig=null, tableContent=
场景片段数量违停车辆数
合计7798
夜间2739
白天2734
隧道2325
), ArticleFig(id=1322949851675242652, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表3, caption=

车辆违停识别数据集不同场景片段数量统计

, figureFileSmall=null, figureFileBig=null, tableContent=
场景片段数量违停车辆数
合计7798
夜间2739
白天2734
隧道2325
), ArticleFig(id=1322949851746545821, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 4, caption=Results of vehicle illegal parking detection under different polynomial orders, figureFileSmall=null, figureFileBig=null, tableContent=
k值准确率/%召回率/%F1分数
168.1478.570.73
288.8989.800.89
394.7992.860.94
490.9191.840.91
588.5486.730.88
), ArticleFig(id=1322949851830431902, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表4, caption=

不同多项式阶数下车辆违停识别结果

, figureFileSmall=null, figureFileBig=null, tableContent=
k值准确率/%召回率/%F1分数
168.1478.570.73
288.8989.800.89
394.7992.860.94
490.9191.840.91
588.5486.730.88
), ArticleFig(id=1322949851897540767, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 5, caption=Experimental results of the proposed algorithm for illegal parking detection, figureFileSmall=null, figureFileBig=null, tableContent=
准确率/%召回率/%F1分数显存/GB帧率/(帧/s)
94.7992.860.942.632
), ArticleFig(id=1322949851964649632, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表5, caption=

违停识别算法实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
准确率/%召回率/%F1分数显存/GB帧率/(帧/s)
94.7992.860.942.632
), ArticleFig(id=1322949852040147105, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 6, caption=Experimental results of highway vehicle tracking, figureFileSmall=null, figureFileBig=null, tableContent=
方法HOTA/%MOTA/%IDF1/%帧率/(帧/s)
Bytetrack62.975.470.8126
BotSORT64.175.868.746
StrongSORT65.477.173.65
Deep OC-SORT62.473.366.46
MambaTrack66.778.877.816
BoostTrack++67.279.278.424
双重注意力66.377.475.352
), ArticleFig(id=1322949852115644578, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表6, caption=

高速公路车辆跟踪实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
方法HOTA/%MOTA/%IDF1/%帧率/(帧/s)
Bytetrack62.975.470.8126
BotSORT64.175.868.746
StrongSORT65.477.173.65
Deep OC-SORT62.473.366.46
MambaTrack66.778.877.816
BoostTrack++67.279.278.424
双重注意力66.377.475.352
), ArticleFig(id=1322949852186947747, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 7, caption=Comparative experimental results of vehicle illegal parking detection, figureFileSmall=null, figureFileBig=null, tableContent=
方法夜间白天隧道平均
P/%R/%F1

帧率

/(帧/s)

P/%R/%F1

帧率

/(帧/s)

P/%R/%F1

帧率

/(帧/s)

P/%R/%F1

显存

/GB

帧率

/(帧/s)

方法165.6353.850.592976.4778.790.782870.3773.080.722870.9770.410.712.128
方法265.7964.100.653184.3881.820.833083.3376.920.803076.6077.550.772.330
方法369.2369.230.694080.5687.880.843780.7780.770.813876.2481.630.792.538
方法474.3674.360.742688.2490.910.902684.6284.620.852781.8286.420.841.827
本文92.1189.740.912893.9496.970.953492.0092.310.923194.7992.860.942.632
), ArticleFig(id=1322949852258250916, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表7, caption=

车辆违停识别对比实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
方法夜间白天隧道平均
P/%R/%F1

帧率

/(帧/s)

P/%R/%F1

帧率

/(帧/s)

P/%R/%F1

帧率

/(帧/s)

P/%R/%F1

显存

/GB

帧率

/(帧/s)

方法165.6353.850.592976.4778.790.782870.3773.080.722870.9770.410.712.128
方法265.7964.100.653184.3881.820.833083.3376.920.803076.6077.550.772.330
方法369.2369.230.694080.5687.880.843780.7780.770.813876.2481.630.792.538
方法474.3674.360.742688.2490.910.902684.6284.620.852781.8286.420.841.827
本文92.1189.740.912893.9496.970.953492.0092.310.923194.7992.860.942.632
), ArticleFig(id=1322949852316971173, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 8, caption=Ablation experiment on DAFM sub-modules, figureFileSmall=null, figureFileBig=null, tableContent=
实验组别HOTAMOTAIDF1
1:Baseline64.175.868.7
2:仅SAM65.876.973.2
3:仅SEBlocks64.776.171.4
4:DAFM66.377.475.3
), ArticleFig(id=1322949852379885734, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表8, caption=

DAFM子模块消融实验

, figureFileSmall=null, figureFileBig=null, tableContent=
实验组别HOTAMOTAIDF1
1:Baseline64.175.868.7
2:仅SAM65.876.973.2
3:仅SEBlocks64.776.171.4
4:DAFM66.377.475.3
), ArticleFig(id=1322949852451188903, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=EN, label=Table 9, caption=Ablation experimental results of highway vehicle illegal parking detection, figureFileSmall=null, figureFileBig=null, tableContent=
方法DAFMAIDTSAP/%R/%F1
基线---71.1675.510.73
方案1√--63.2380.610.71
方案2√√-80.9090.820.86
本文√√√94.7992.860.94
), ArticleFig(id=1322949852518297768, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, language=CN, label=表9, caption=

高速车辆违停识别消融实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
方法DAFMAIDTSAP/%R/%F1
基线---71.1675.510.73
方案1√--63.2380.610.71
方案2√√-80.9090.820.86
本文√√√94.7992.860.94
)], attaches=null, journal=Journal(id=1249023527618129992, delFlag=0, nameCn=中国图象图形学报, nameEn=Journal of Image and Graphics, nameHistory1=null, nameHistory2=null, issn=1006-8961, eissn=null, cn=11-3758, coden=CODEN ZTTXFZ, periodic=0, 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=uirXtX858YS3zEpFXZttJA==, journalPrice=null, startedYear=null, abbrevIsoEn=Journal of Image and Graphics, journalRemark=null, publicationField=null, createdTime=1775720014721, updatedTime=1788951488513, createdBy=18614031015, updatedBy=13041195026, firstLetterCn=Z, firstLetterEn=Z, subjectCode=Electronic Information, subjectName=null, subjectCodeEn=Electronic Information, subjectNameEn=null, picCn=uirXtX858YS3zEpFXZttJA==, picEn=bud7qaxfvWHeFsbyBTAiKQ==, jcr=null, cjcr=null, exts=[JournalExt(id=1304520351199814396, language=CN, name=中国图象图形学报, 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=1788951488542, updatedTime=1788951488542, createdBy=13041195026, updatedBy=13041195026, submissionGuidelinesUrl=, submissionAuthorUrl=https://journal.ids.fzyun.cn/auth/realms/journal/protocol/openid-connect/auth?client_id=journal-cjig-author&redirect_uri=https%3A%2F%2Fcjig.portal.founderss.cn%2Foauth%2Fcallback&response_type=code&scope=phone+openid+email+profile&state=e6369def-2842-41d8, submissionEditorUrl=https://journal.ids.fzyun.cn/auth/realms/journal/protocol/openid-connect/auth?client_id=journal-cjig-editor&redirect_uri=https%3A%2F%2Fcjigeditor.portal.founderss.cn%2Foauth%2Fcallback&response_type=code&scope=phone+openid+email+profile&state=9ccec05b-6bd, submissionReviewUrl=https://journal.ids.fzyun.cn/auth/realms/journal/protocol/openid-connect/auth?client_id=journal-cjig-author&redirect_uri=https%3A%2F%2Fcjig.portal.founderss.cn%2Foauth%2Fcallback&response_type=code&scope=phone+openid+email+profile&state=1e8a31c8-5434-4f78, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1304520351250146045, language=EN, name=Journal of Image and 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=1788951488554, updatedTime=1788951488554, createdBy=13041195026, updatedBy=13041195026, submissionGuidelinesUrl=, submissionAuthorUrl=https://journal.ids.fzyun.cn/auth/realms/journal/protocol/openid-connect/auth?client_id=journal-cjig-author&redirect_uri=https%3A%2F%2Fcjig.portal.founderss.cn%2Foauth%2Fcallback&response_type=code&scope=phone+openid+email+profile&state=e6369def-2842-41d8, submissionEditorUrl=https://journal.ids.fzyun.cn/auth/realms/journal/protocol/openid-connect/auth?client_id=journal-cjig-editor&redirect_uri=https%3A%2F%2Fcjigeditor.portal.founderss.cn%2Foauth%2Fcallback&response_type=code&scope=phone+openid+email+profile&state=9ccec05b-6bd, submissionReviewUrl=https://journal.ids.fzyun.cn/auth/realms/journal/protocol/openid-connect/auth?client_id=journal-cjig-author&redirect_uri=https%3A%2F%2Fcjig.portal.founderss.cn%2Foauth%2Fcallback&response_type=code&scope=phone+openid+email+profile&state=1e8a31c8-5434-4f78, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1249024232475115590, websiteList=[Website(id=1249025782459334881, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1249024232475115590, 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/zgtxtxxb/CN, language=CN, createTime=1775720552315, createBy=18614031015, updateTime=1775720586268, updateBy=18614031015, name=中国图象图形学报-中文, tplId=1146099689490845704, title=中国图象图形学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1249026166254928133, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=articleTextType, value=kx, createTime=1775720643819, updateTime=1775720643819, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166221373698, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=banner, value=null, createTime=1775720643811, updateTime=1775720643811, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166271705352, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=grayFlag, value=0, createTime=1775720643823, updateTime=1775720643823, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166212985089, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=logo, value=https://castjournals.cast.org.cn/joweb/zgtxtxxb/CN/file/pic?fileId=TDRjKTHfgAnvFKZaDA70wA==, createTime=1775720643809, updateTime=1775720643809, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166288482570, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=minRunFlag, value=0, createTime=1775720643827, updateTime=1775720643827, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166246539524, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/zgtxtxxb/CN/file/pic, createTime=1775720643817, updateTime=1775720643817, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166280093961, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=silenceFlag, value=0, createTime=1775720643825, updateTime=1775720643825, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166233956611, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1775720643814, updateTime=1775720643814, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166259122438, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=themeColor, value=null, createTime=1775720643820, updateTime=1775720643820, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026166267511047, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782459334881, code=themeStyle, value=null, createTime=1775720643822, updateTime=1775720643822, creator=18614031015, updator=18614031015)]), Website(id=1249025782681633001, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1249024232475115590, 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/zgtxtxxb/EN, language=EN, createTime=1775720552368, createBy=18614031015, updateTime=1775720607118, updateBy=18614031015, name=中国图象图形学报-英文, tplId=1146101810881728533, title=Journal of Image and Graphics, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1249026195371786511, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=articleTextType, value=kx, createTime=1775720650761, updateTime=1775720650761, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195355009292, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=banner, value=null, createTime=1775720650757, updateTime=1775720650757, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195392758034, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=grayFlag, value=0, createTime=1775720650766, updateTime=1775720650766, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195342426379, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=logo, value=https://castjournals.cast.org.cn/joweb/zgtxtxxb/EN/file/pic?fileId=TDRjKTHfgAnvFKZaDA70wA==, createTime=1775720650754, updateTime=1775720650754, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195409535252, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=minRunFlag, value=0, createTime=1775720650770, updateTime=1775720650770, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195367592206, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/zgtxtxxb/EN/file/pic, createTime=1775720650760, updateTime=1775720650760, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195401146643, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=silenceFlag, value=0, createTime=1775720650768, updateTime=1775720650768, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195359203597, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1775720650758, updateTime=1775720650758, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195380175120, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=themeColor, value=null, createTime=1775720650763, updateTime=1775720650763, creator=18614031015, updator=18614031015), WebsiteProps(id=1249026195388563729, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1249025782681633001, code=themeStyle, value=null, createTime=1775720650765, updateTime=1775720650765, creator=18614031015, updator=18614031015)])], journalTitle=中国图象图形学报, weixinUrl=null, journalUrl=https://www.cjig.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Journal of Image and Graphics, journalPhotoCn=uirXtX858YS3zEpFXZttJA==, journalPhotoEn=bud7qaxfvWHeFsbyBTAiKQ==, journalFirstLetter=Z, 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, interPubPlatform=, interPubPlatformUrl=null), detailUrlCn=https://castjournals.cast.org.cn/joweb/zgtxtxxb/CN/10.11834/jig.250354, detailUrlEn=https://castjournals.cast.org.cn/joweb/zgtxtxxb/EN/10.11834/jig.250354, pdfUrlCn=https://castjournals.cast.org.cn/joweb/zgtxtxxb/CN/PDF/10.11834/jig.250354, pdfUrlEn=https://castjournals.cast.org.cn/joweb/zgtxtxxb/EN/PDF/10.11834/jig.250354, aliStartDate=0, aliEndDate=0, collectionFlag=false, citedCount=null, citedUrl=null, previewStatus=0, delFlag=0, hasFullText=1, orderTime=1784131200000, fullTextJson=null, articleText=null, reference=null)
收藏切换
融合双重注意力特征与时空特征的高速公路违停识别
收藏切换
PDF下载
刘电 1 , 尹娜 1 , 张泽宁 2 , 吕洪燕 2 , 罗旭彬 2 , 王昊翔 1, *
中国图象图形学报 | 图像分析和识别 2026,31(7): 2614-2630
收起
收藏切换
中国图象图形学报 |图像分析和识别 2026 , 31 (7) : 2614 -2630
融合双重注意力特征与时空特征的高速公路违停识别
全屏
[Author(id=1322949842305167440, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=csdianl@mail.scut.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949842393247826, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842305167440, language=EN, stringName=Dian Liu, firstName=Dian, middleName=null, lastName=Liu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949842485522515, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842305167440, language=CN, stringName=刘电, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 华南理工大学计算机科学与工程学院,广州 510006, bio={"content":"

刘电,男,博士研究生,主要研究方向为深度学习和目标检测与语义分割。E-mail:

"}, bioImg=null, bioContent=

刘电,男,博士研究生,主要研究方向为深度学习和目标检测与语义分割。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949840627445833, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=1, ext=[AuthorCompanyExt(id=1322949840635834442, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China), AuthorCompanyExt(id=1322949840648417355, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 华南理工大学计算机科学与工程学院,广州 510006)])]), Author(id=1322949842561019989, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=202220143482@mail.scut.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949842653294679, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842561019989, language=EN, stringName=Na Yin, firstName=Na, middleName=null, lastName=Yin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949842724597848, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842561019989, language=CN, stringName=尹娜, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 华南理工大学计算机科学与工程学院,广州 510006, bio={"content":"

尹娜,女,硕士研究生,主要研究方向为计算机视觉和目标检测。E-mail:

"}, bioImg=null, bioContent=

尹娜,女,硕士研究生,主要研究方向为计算机视觉和目标检测。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949840627445833, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=1, ext=[AuthorCompanyExt(id=1322949840635834442, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China), AuthorCompanyExt(id=1322949840648417355, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 华南理工大学计算机科学与工程学院,广州 510006)])]), Author(id=1322949842833649754, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=zznmizzou@gmail.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949842925924444, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842833649754, language=EN, stringName=Zening Zhang, firstName=Zening, middleName=null, lastName=Zhang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949843001421917, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949842833649754, language=CN, stringName=张泽宁, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 广州国交润万交通信息有限公司,广州 510665, bio={"content":"

张泽宁,男,软件工程师,主要研究方向为计算机视觉、自然语言处理、生成式模型以及多模态学习。E-mail:

"}, bioImg=null, bioContent=

张泽宁,男,软件工程师,主要研究方向为计算机视觉、自然语言处理、生成式模型以及多模态学习。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949842191921228, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=2, ext=[AuthorCompanyExt(id=1322949842200309837, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China), AuthorCompanyExt(id=1322949842212892750, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 广州国交润万交通信息有限公司,广州 510665)])]), Author(id=1322949843261468767, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=lhy@runonetech.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949843345354849, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949843261468767, language=EN, stringName=Hongyan Lyu, firstName=Hongyan, middleName=null, lastName=Lyu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949843433435234, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949843261468767, language=CN, stringName=吕洪燕, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 广州国交润万交通信息有限公司,广州 510665, bio={"content":"

吕洪燕,男,正高级工程师,主要研究方向为智能交通、交通管控和深度学习。E-mail:

"}, bioImg=null, bioContent=

吕洪燕,男,正高级工程师,主要研究方向为智能交通、交通管控和深度学习。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949842191921228, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=2, ext=[AuthorCompanyExt(id=1322949842200309837, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China), AuthorCompanyExt(id=1322949842212892750, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 广州国交润万交通信息有限公司,广州 510665)])]), Author(id=1322949843508932708, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=515664321@qq.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1322949843739619430, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949843508932708, language=EN, stringName=Xubin Luo, firstName=Xubin, middleName=null, lastName=Luo, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949843815116903, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949843508932708, language=CN, stringName=罗旭彬, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 广州国交润万交通信息有限公司,广州 510665, bio={"content":"

罗旭彬,男,软件工程师,主要研究方向为计算机视觉、自然语言处理和模式识别。E-mail:

"}, bioImg=null, bioContent=

罗旭彬,男,软件工程师,主要研究方向为计算机视觉、自然语言处理和模式识别。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949842191921228, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=2, ext=[AuthorCompanyExt(id=1322949842200309837, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China), AuthorCompanyExt(id=1322949842212892750, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949842191921228, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 广州国交润万交通信息有限公司,广州 510665)])]), Author(id=1322949844087746665, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, orderNo=5, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=hxwang@scut.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1322949844171632747, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949844087746665, language=EN, stringName=Haoxiang Wang, firstName=Haoxiang, middleName=null, lastName=Wang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, *, address=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1322949844263907436, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, authorId=1322949844087746665, language=CN, stringName=王昊翔, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, *, address=1 华南理工大学计算机科学与工程学院,广州 510006, bio={"content":"

王昊翔,通信作者,男,副教授,硕士生导师,主要研究方向为机器视觉与模式识别。E-mail:

"}, bioImg=null, bioContent=

王昊翔,通信作者,男,副教授,硕士生导师,主要研究方向为机器视觉与模式识别。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322949840627445833, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, xref=1, ext=[AuthorCompanyExt(id=1322949840635834442, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China), AuthorCompanyExt(id=1322949840648417355, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322949834885443622, companyId=1322949840627445833, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 华南理工大学计算机科学与工程学院,广州 510006)])])]
刘电1 , 尹娜1 , 张泽宁2 , 吕洪燕2 , 罗旭彬2 , 王昊翔1, *
作者信息
  • 1 华南理工大学计算机科学与工程学院,广州 510006
  • 2 广州国交润万交通信息有限公司,广州 510665
通讯作者:
作者简介:

刘电,男,博士研究生,主要研究方向为深度学习和目标检测与语义分割。E-mail:

尹娜,女,硕士研究生,主要研究方向为计算机视觉和目标检测。E-mail:

张泽宁,男,软件工程师,主要研究方向为计算机视觉、自然语言处理、生成式模型以及多模态学习。E-mail:

吕洪燕,男,正高级工程师,主要研究方向为智能交通、交通管控和深度学习。E-mail:

罗旭彬,男,软件工程师,主要研究方向为计算机视觉、自然语言处理和模式识别。E-mail:

王昊翔,通信作者,男,副教授,硕士生导师,主要研究方向为机器视觉与模式识别。E-mail:

Highway parking violation detection by integrating dual-attention and spatiotemporal features
Dian Liu1 , Na Yin1 , Zening Zhang2 , Hongyan Lyu2 , Xubin Luo2 , Haoxiang Wang1, *
Affiliations
  • 1 School of Computer Science and Engineering,South China University of Technology,Guangzhou 510006,China
  • 2 Guangzhou Runone Traffic Information Co.,Ltd.,Guangzhou 510665,China
出版时间: 2026-07-16 doi: 10.11834/jig.250354
文章导航
收藏切换
目的

为提升高速公路交通管理效率与道路安全,提出一种面向监控视频的违法停车自动识别方法,以解决车流密集场景中目标遮挡、摄像机透视畸变及夜间光照干扰导致的身份混淆与误判问题。

方法

引入了双重注意力特征融合结构,增强YOLOv8(you only look once)提取的多尺度特征的表达能力,并将生成的外观特征用于后续的轨迹匹配阶段,从而在保持高效跟踪性能的同时,有效缓解了目标身份混淆与切换的现象。此外,还设计了一种基于帧间位移自适应的判别机制来识别违停行为,有效降低了由视角畸变和边框抖动造成的误判和漏判风险。针对夜间检测的困难,进一步引入基于时间序列信号分析的危险报警闪光灯检测技术,以提升夜间判定的稳定性与准确性,从而增强整体算法的鲁棒性。

结果

鉴于当前尚无公开可用的高速公路违停检测数据集,结合实际交通管理需求,自主构建覆盖多场景的专用数据集,并在此基础上进行实验验证。结果表明,所提方法能够良好适应不同时间段及复杂条件下的高速公路场景,违停识别准确率达94.79%,召回率达92.86%,F1分数为0.94,显著优于对比方法。所提方法的图像处理速度为32帧/s,满足实时处理需求。

结论

提出的融合双重注意力特征与时空特征的方法,在高速公路违停识别上成效显著,该研究具备显著的开创性和实践指导意义。

高速公路  /  时空特征  /  注意力特征  /  违停识别  /  双重注意力
Objective

Automatic detection of illegal parking on highways through surveillance video is a crucial task for modern traffic management systems because it directly affects traffic flow,public safety,and operational efficiency. However,detecting parking violations in highway environments presents substantial challenges because of complex traffic dynamics,varying lighting conditions,and broad surveillance coverage. These challenges include frequent vehicle occlusions in dense traffic that lead to identity confusion and incorrect tracking,substantial perspective distortion that causes slow-moving vehicles to be mistakenly identified as parked,and degraded detection performance under low-illumination conditions,where vehicle features become blurred and are prone to false detection because of environmental lighting effects.

Method

To address these problems,this study proposes a novel highway illegal parking detection algorithm that integrates dual-attention feature fusion and spatiotemporal behavior analysis. The approach introduces a dual-attention fusion module to enhance multiscale feature representations extracted from YOLOv8,which are used as appearance descriptors in a vehicle tracking framework on the basis of BotSORT-ReID. This framework directly leverages YOLOv8’s multiscale feature outputs and enhances them with attention mechanisms in spatial and channel dimensions. The spatial attention module emphasizes critical regions related to vehicle targets,suppressing background noise,whereas the channel attention mechanism selectively emphasizes discriminative feature channels relevant to object identity. These enhancements substantially improve the robustness and accuracy of vehicle tracking,especially in scenarios involving occlusion,background clutter,or varying viewing angles. The second component of the algorithm focuses on violation judgment through spatiotemporal feature accumulation. Specifically,it uses an adaptive interframe displacement estimation method on the basis of exponentially weighted moving averages to identify low-speed or stationary vehicles over a sequence of frames. A normalized displacement model is employed to correct perspective distortions across different regions of the image,ensuring that movement estimates are reliable even when vehicle scale varies with the distance from the camera. This method smooths short-term fluctuations in object position and helps avoid misclassification caused by detection box jitter or gradual deceleration. A secondary validation mechanism is proposed to address the complexity of nighttime detection,especially in environments with poor visibility or interference from artificial lights. This process involves analyzing the temporal characteristics of brightness changes to identify flashing hazard lights,which are a strong indicator of stationary vehicles in emergency situations. Time-series analysis is conducted on grayscale intensity values extracted from target bounding boxes across frames. Preprocessing steps,including Savitzky-Golay filtering,detrending,and temporal gradient computation,are applied to isolate periodic brightness variations. Subsequently,Welch’s method is used for power spectral density estimation,and dominant frequencies are extracted and compared with known hazard light flashing frequency ranges. Only when these frequencies match expected patterns is the object confirmed to be illegally parked,thereby reducing false positives from nonvehicle light sources or reflections.

Result

Given the absence of publicly available datasets for highway parking violation detection,three dedicated highway datasets are constructed for object detection,vehicle tracking,and parking violation recognition. The datasets consist of 8 532 labeled images for detection,34 600 frames for tracking,and 77 annotated video sequences (average over 1 min,covering 98 illegal parking vehicles) for violation recognition. Moreover,they cover diverse conditions,including daytime,nighttime,tunnels,and adverse weather (e.g.,rain and fog),thereby ensuring comprehensive evaluation across complex highway scenarios. Experimental results on the constructed datasets show that the proposed algorithm consistently outperforms several benchmark methods in terms of precision,recall,and F1 score. Notably,dual-attention feature extraction improves identity consistency in vehicle tracking,thereby reducing ID switches and increasing tracking accuracy. The adaptive displacement method demonstrates strong resistance to jitter and perspective artifacts,and the hazard light detection module substantially reduces nighttime false positives and improves robustness under difficult lighting. Ablation studies further confirm the contribution of each component. For instance,removing the dual-attention module results in reduced tracking quality and an increase in identity confusion. Without the adaptive displacement method,the system becomes highly vulnerable to misclassification because of minor motion or camera instability. Meanwhile,excluding the temporal signal-based hazard light detection leads to a rise in false alarms during nighttime evaluations. The full system achieves a balanced trade-off between high precision and high recall,demonstrating its effectiveness in real-world applications.

Conclusion

Overall,the proposed algorithm presents a unified,efficient,accurate solution for illegal parking detection on highways. By combining deep learning-based object detection,attention-guided feature fusion,motion pattern recognition,and signal-based validation,this algorithm addresses the key limitations of current methods and achieves robust performance across varying highway surveillance conditions. The integration of lightweight attention modules into the YOLOv8 backbone and the design of spatiotemporal behavioral analysis allow for real-time processing while maintaining high accuracy. The system is thus well-suited for deployment in intelligent transportation systems and smart city infrastructure where reliability and adaptability are paramount. The methodology also provides a scalable framework for extending similar detection strategies to other traffic violations or abnormal behaviors in complex road environments.

highway  /  spatiotemporal feature  /  attention feature  /  parking violation detection  /  dual attention
刘电, 尹娜, 张泽宁, 吕洪燕, 罗旭彬, 王昊翔. 融合双重注意力特征与时空特征的高速公路违停识别. 中国图象图形学报, 2026 , 31 (7) : 2614 -2630 . DOI: 10.11834/jig.250354
Dian Liu, Na Yin, Zening Zhang, Hongyan Lyu, Xubin Luo, Haoxiang Wang. Highway parking violation detection by integrating dual-attention and spatiotemporal features[J]. Journal of Image and Graphics, 2026 , 31 (7) : 2614 -2630 . DOI: 10.11834/jig.250354
违法停车是常见的交通违法行为之一,不仅会造成交通拥堵,干扰正常的通行秩序,还可能引发交通事故,增加道路安全风险。尤其在高速公路环境下,违停现象带来的危害更加严重,极大地威胁到公众的生命和财产安全。因此,有效检测高速公路车辆违停行为成为智能交通系统研究中的一个核心问题。早在20世纪90年代,该领域便引起了学术界的广泛关注,相关研究逐步展开,并随着深度学习技术的兴起而进入快速发展阶段。
在车辆违停识别的相关研究中,大多数方法基于图像或视频数据进行分析,部分研究结合传感器信息、全球定位系统(global positioning system,GPS)或轨迹数据等多源信息,以提升检测的准确性。虽然与视觉数据相比,传感器和GPS等非视觉信息对光照、气候等外部条件的依赖较小,因此在检测稳定性方面具有一定优势,但这些方式往往需要额外的硬件设备或依赖外部系统,因而在实际应用中受到成本和部署条件的限制,难以大规模推广。相较而言,基于视频和图像的检测方法由于监控设备的普遍部署以及图像处理技术的不断成熟,依然是当前最为常见和实用的技术路径。因此,本研究依托高速公路的监控视频与图像数据开展违停行为的识别研究。
本文立足于高速公路视频监控场景,提出一种融合双重注意力机制与时空特征的违停识别方法。主要的创新点包括以下方面:1)提出一种融合空间注意力与通道注意力的双重注意力特征融合模块,在保持轻量化的同时显著提升了多目标跟踪中的身份稳定性;2)设计自适应帧间位移机制,有效解决了透视畸变和检测框抖动导致的误判问题;3)引入基于时序信号分析的危险报警闪光灯检测方法,显著提高了夜间与低照度场景下的违停识别准确性。
同时,构建了面向高速公路的车辆违停数据集,并开展了多组对比实验。实验结果表明,该算法在适应不同时间段与天气条件方面展现出良好的鲁棒性与泛化能力。值得指出的是,当前在公开研究资源中,尚无专门面向高速公路违停行为的标准化数据集。已有研究多聚焦于城市道路或停车场场景,难以覆盖高速公路环境中监控视角远、目标尺度变化大以及违停样本稀疏等特点。为此,本文首次构建了面向高速公路的违停检测专用数据集,包括目标检测、跟踪和违停行为识别等多个任务维度,覆盖昼夜时段、隧道场景以及多种气象条件,填补了该领域数据资源的空白,该数据集的构建为违停检测任务提供了高质量的数据支撑。
早期的违停识别依赖传统图像处理技术,例如Hassan等人(2012)提出的两阶段识别框架,利用前景分割和自适应跟踪有效识别静止车辆,克服了光照变化和遮挡的干扰。Wahyono等人(2016)提出了一种基于双背景模型的累积前景差分方法,利用具有不同学习率的短期与长期高斯混合模型提取前景像素,通过连通域标记形成静态区域,并结合基于规则的筛选与检测跟踪机制,实现对违停车辆的稳健识别。
尽管基于传统图像处理技术的方法验证了计算机视觉在智能交通领域的应用潜力,并为违停检测提供了基础的技术框架,但这些方法仍存在一些问题,如对特征的过度依赖人工设计、在不同场景下的泛化能力不足以及准确率的限制等。
随着科技的不断进步,传统图像处理方法在应对复杂场景时逐渐显露出局限性,机器学习凭借其强大的自适应能力和数据处理能力,逐渐成为研究的焦点。后续研究者逐渐将支持向量机(support vector machine,SVM)、随机森林(random forest,RF)等传统机器学习算法引入到违法停车检测任务中。例如,Ketcham等人(2019)提出了一种融合SVM与线性回归的检测方法,通过非静态背景差分技术提取车辆的前景区域,随后利用方向梯度直方图(histogram of oriented gradients,HOG)对图像中车辆特征进行提取,最后借助SVM进行分类判断,并通过线性回归进一步优化检测结果。Matsuda等人(2021)将车辆的边界框属性、极值、平均值和方差等统计特征,以及速度与加速度等动态特征作为输入,训练一个基于随机森林的分类模型,结合逻辑回归方法,对路边停车和非道路区域停车的比例进行平衡处理,从而识别违停车辆。
尽管将机器学习方法应用于违停识别在一定程度上提高了算法的鲁棒性和泛化能力,但在面临如小目标检测、车辆遮挡等复杂实际场景时,仍难以克服性能瓶颈,表现仍有待进一步提升。
随着数据量的爆炸式增长和任务复杂度的提升,深度学习以其多层非线性变换的强大特征学习能力,开启该领域研究的新篇章。近年来,研究者们将注意力转向利用深度神经网络提升车辆违停识别的准确性与处理效率。诸多研究尝试将YOLOv3(you only look once)(Peng等,2022)、YOLOv8(Park等,2025)和SSD(single shot multibox detector)(Xie等,2017)等高效目标检测算法应用于违停场景以满足实时性需求。
其中,Hamzah等人(2022)提出基于Faster R-CNN(faster region-based convolutional neural network)(Ren等,2017)的违停检测思路,通过4阶段卷积神经网络提取图像特征,随后由Inception V2(Ioffe和Szegedy,2015)进行进一步处理,最终通过Softmax分类器与回归模块确定车辆位置及其违停状态。汪琦等人(2024)整合多个域的信息,并采用共享子空间度量学习机制,有效解决了跨域车辆特征不一致的问题,提升了车辆重识别精度。余烨等人(2023)通过引入深度卷积网络和残差学习机制,有效提升了夜间低光照条件下车辆图像的质量,增强了车辆特征的可见性,特别是在车型识别中,能够提升车辆的精确识别率。
与此同时,基于Transformer的检测架构,如检测变换器(detection Transformer,DETR)及其改进模型也引入到交通场景目标检测和时空特征建模中,利用全局自注意力机制提升了长距离依赖建模能力,在遮挡与拥挤环境下展现出更高的鲁棒性。Wang等人(2025)在轻量化检测框架中引入高效注意力模块,在保证实时性的同时保持较高精度,为资源受限的道路监控设备提供了可行性方案。Zeng等人(2022)通过引入跨帧交互注意力机制,提升了在交通监控中对车辆轨迹和行为的建模能力。徐晗等人(2025)利用Transformer的全局建模能力并引入上下文感知注意力机制,通过自注意力和交叉注意力层融合模板与搜索区域特征,提升复杂场景下的跟踪鲁棒性和准确性。这些研究展示了注意力机制在提升检测准确性和适应性方面的潜力。
综上,已有研究在城市道路与普通场景中取得了进展,但在高速公路场景中仍面临监控范围广、车辆遮挡严重以及低照度干扰明显等挑战。尤其是缺乏公开可用的高速公路违停数据集,制约了算法的训练与验证。这些不足凸显了开展针对高速公路违停识别研究的必要性。
本文提出的高速公路车辆违停识别算法的整体流程如图1所示。算法包含两个核心功能模块:基于双重注意力特征融合的轨迹识别模块和基于时空特征的违停判别模块。
轨迹识别模块选用YOLOv8为BotSORT-ReID跟踪算法(Aharon等,2022)的目标检测组件,并将YOLOv8输出的特征通过双重注意力特征融合模块(dual-attention fusion module,DAFM)处理,加强特征的表达能力。融合过程有助于优化BotSORT-ReID跟踪算法中的轨迹关联环节。
违停判别模块包含两个子模块:自适应帧间位移(adaptive inter-frame displacement,AID)的初级判定子模块和基于时序信号分析(time series analysis,TSA)的次级验证子模块。初级验证子模块通过计算每条轨迹的自适应帧间位移来判别其运动状态,如果是低速运动或静止状态,则进入次级验证子模块结合目标类型进行判别。次级验证子模块对暗光疑似车辆类型进行报警闪光灯时序信号分析,降低夜间干扰目标误报;对其他类型进行归一化帧间位移判别目标静止状态,提高违停算法在不同场景下的鲁棒性。
BotSORT-ReID(Aharon等,2022)通过融合外观特征与运动信息,实现多目标追踪的高鲁棒性与准确性。轨迹识别模块采用YOLOv8作为检测模块,基于BotSORT-ReID算法对车辆进行检测与追踪,获取其轨迹用于后续的违停行为识别。不同于原始方法中使用的特征提取模块,本研究直接利用YOLOv8所输出的特征图,并引入一种双重注意力特征融合机制以实现特征的提取与匹配。
YOLOv8在颈部网络输出3种尺度的特征图,经过解耦头处理后,这些特征在尺度、粒度及关注维度上各具差异,其所表达的特征重点亦不尽相同。为了更高效地融合这些多样化的特征输出,并增强模型在重识别任务中的特征表达能力,提出了一种融合空间注意力与通道注意力的双重注意力特征融合模块(DAFM),结构如图2所示,包含以下4个核心阶段:
1)空间注意力融合阶段。3种不同尺度的特征分别输入空间注意力融合子模块(spatial attention fusion module,SAFM),生成增强后的特征图。利用空间注意力机制(spatial attention module,SAM)(Woo等,2018),动态学习特征图中各空间位置与当前任务的关联性,对关键区域赋予更高权重,使模型专注于重识别最相关的空间特征。
2)特征拼接阶段。将上述3个增强特征图在通道维度上进行拼接,通过此方式,来自不同解耦头的特征信息得以完整保留与整合,提升特征表达的多样性以及对不同视角特征的兼容性。
3)通道注意力增强阶段。由于拼接操作使得特征图的通道数增加,而各通道在任务中的作用并不一致,因此引入SEBlocks(Hu等,2020)实现通道注意力增强。这一机制可聚焦于对重识别具有判别力的通道,同时抑制无关或低价值信息,从而强化模型的识别能力。该结构详见图3。
4)表观特征生成阶段。为进一步挖掘融合特征中的局部判别信息,将融合注意力机制后的特征送入一个3 × 3的卷积层以提取更加精细的特征细节,并通过平均池化操作进行降维,生成最终用于轨迹关联的特征向量。
对比当前的主流注意力机制,SENet(Hu等,2020)与CBAM(Woo等,2018)均为通用目标检测设计,未适配高速公路场景特性,DAFM的差异主要体现在以下方面:
1)注意力作用维度不同。SENet仅关注通道维度,通过全局池化与全连接层分配通道权重筛选有效特征;CBAM采用“先进行通道注意力处理,再开展空间注意力处理”的串行流程,筛选通道再聚焦空间区域;而DAFM则采用“先进行空间注意力增强,再开展通道注意力筛选”的组合逻辑,先对多尺度特征分别做空间增强,抑制遮挡与背景,再拼接后做通道筛选,实现双维度协同优化。
2)多尺度特征处理方式不同。SENet采用单尺度输入,没有专门的多尺度处理设计,容易丢失远处小目标特征;CBAM对单尺度特征图进行注意力增强,并且其空间注意力通过7 × 7固定卷积核生成权重,感受野固定;DAFM则专门适配高速公路多尺度车辆场景,对YOLOv8输出的1/8、1/16、1/32这3种尺度特征分别进行空间注意力增强,空间注意力基于全局池化动态学习权重,可实时聚焦车辆未遮挡区域,提升抗遮挡能力。
综上所述,双重注意力特征融合模块直接利用YOLOv8的原生特征输出,无须额外图像特征提取,有效降低了系统开销并提升追踪效率。同时,通过结合空间与通道注意力机制,有效减轻了背景杂乱、遮挡及姿态变化等对特征提取造成的不利影响,增强了模型对目标区域的感知能力。
基于时空特征的违停判别模块的详细流程如图4所示。其中,F(j )表示车辆在第j帧是否处于静止状态,1和0分别表示静止状态和运动状态。模块通过自适应帧间位移方法对车辆轨迹数据进行初步分析,识别出低速运动或静止的目标,即违停候选车辆。随后,针对YOLOv8识别出的夜间低光条件下的疑似目标,应用基于时序信号分析的危险报警闪光灯检测技术,进一步过滤可能的误报情况。
在违法停车检测任务中,精准判断目标的移动状态是判定车辆是否存在违停行为的关键。然而在实际应用中,由于目标检测框可能出现轻微抖动,以及车辆与摄像机之间距离的变化会造成透视畸变,直接利用相邻帧之间的位移来判断违停容易受到干扰,引入误差。为了解决这一问题,提出了一种基于指数加权移动平均(exponentially weighted moving average,EWMA)策略的自适应帧间位移计算方法。
1)归一化帧间位移。为了克服由于车辆与摄像头之间距离差异所造成的透视畸变问题,针对每一个具体的道路场景,对监控视频中车辆的运动轨迹进行了持续一段时间的采集与分析,从而获得该场景下车辆在连续帧之间的位移数据分布情况,如图5所示。图5中的纵轴反映了车辆在不同空间位置上的帧间位移幅度。根据这些数据拟合出一条标准的帧间位移参考曲线,其数学表达形式为
f(x)=∑k=0Kakxk
式中,x表示轨迹框中心的纵向坐标,
f(x)
函数表示位移的拟合表达式,k为所使用的多项式拟合的阶数,
αk
表示对应k次幂的系数值,这些系数可以通过最小二乘法对目标函数进行优化求解以获得。
为了更有效地捕捉低速甚至静止状态下车辆的轨迹变化,引入了权重因子的加权目标函数,其形式为
minwup⋅∑i=1Nupyi-∑k=0Kakxik2+wdown⋅∑j=1Ndownyj-∑k=0Kakxjk2
式中,
xi
,
yi
表示所采集的轨迹中心点的纵向坐标及其对应的帧间位移值;
wup
和
wdown
分别是分配给位于拟合曲线上方与下方的轨迹点的权重参数,而
Nup
和
Ndown
则表示这两类轨迹点的数量。如图5所示,越靠近顶端即纵坐标越小,目标的帧间位移区间越小,运动状态越难判别。通过合理设置
wup
和
wdown
的值,可以使拟合曲线趋向于靠近轨迹点的下方,有助于更精准地识别车辆的细微位移,尤其是对于远处低速移动或接近静止的目标车辆更为敏感。
针对第
i
帧所对应的车辆轨迹
Ti
,其中心点坐标记做
xi
。该车辆在该帧的归一化帧间位移
dN(Ti)
可定义为
dN(Ti)=disp(Ti)f(xi)
式中,
disp(Ti)
表示当前车辆目标在相邻帧之间的实际移动距离,而
f(xi)
代表该位置对应的标准参考帧间位移。通过使用归一化后的帧间位移值判断车辆是否发生移动,能够有效抵消由于摄像头的安装高度、拍摄角度以及镜头焦距等因素所引起的透视畸变影响,从而使车辆移动的判定可以在统一的位移尺度下进行,提升计算的一致性与准确性。
2)自适应帧间位移。在归一化位移的基础之上,引入了指数加权移动平均(EWMA)的思想,提出了自适应帧间位移的更新机制,其计算为
daTi=γdaTi-1+1-γdnTi
式中,
daTi
为当前轨迹在当前第i帧的自适应帧间位移,
dNTi-1
为轨迹在上一帧的自适应帧间位移,
dTi
为轨迹的归一化帧间位移。参数
γ
为平滑因子,控制当前的帧间位移与上一帧的自适应帧间位移的权重,具体定义为
γ=nSLTinSLTi-1+dTi+λ
式中,
nSL(Ti)
表示从目标开始处于低速或静止状态以来,连续处于该状态的帧数。
λ
为一个固定值,用做轻微平滑的参数,避免在特殊情况下分母接近零导致数值不稳定问题。通过动态调整归一化帧间位移与基于历史轨迹信息计算的自适应帧间位移之间的权重分配,能有效应对在实际检测过程中出现的各种问题。当车辆长时间保持静止或低速状态时,平滑因子
γ
将趋近于1,使得更新时更依赖历史轨迹数据,抑制短时间内因检测框不稳定带来的误差。一旦车辆快速启动并造成帧间位移急剧增加,平滑因子
γ
会迅速减小,使当前的归一化位移在更新中占据主导地位,能快速反映车辆状态的变化,避免将其误判为违停。
综上所述,引入自适应帧间位移机制识别车辆的运动状态,能根据不同的动态特性自适应调整策略,显著提升了违停识别的鲁棒性和适应性。
在夜间环境中,由于光照条件较差,车辆的检测准确性容易受到影响,可能出现漏检或误检的情况。为提升夜间车辆的识别能力,引入了“暗光疑似车辆”这一新类别。该策略虽然在一定程度上提高了车辆的检出率,但也引发了新的问题,如路灯、反光标志灯等强光背景容易被误判为车辆。为避免此类误判进一步导致违停检测错误,提出了一种基于时间序列信号分析的二次验证机制。
车辆的双闪警示灯是判断车辆处于特殊停放状态的重要特征,具备显著的可识别性和典型的时间变化特征,在特定的环境下能够有效地区分车辆和背景。通过检测车辆危险报警闪光灯的开启状态,可以在夜间或低能见度条件下起到辅助识别的效果。鉴于危险报警闪光灯的点亮是一种动态过程,仅依赖单帧图像难以准确判断其状态。为此,提出了一种基于时间序列分析的危险报警闪光灯识别方法,通过对车辆目标区域亮度变化的周期性特征进行功率谱分析,提取关键频率成分,实现对车辆危险报警闪光灯闪烁频率的精确识别。
如图6所示,该方法的实现过程包括两个阶段。亮度曲线的预处理阶段从历史帧的轨迹序列中提取波形信号进行预处理,曲线频谱分析阶段对预处理后的波形信号进行分析,提取主要的频率信息作为危险报警闪光灯检测的依据。
1)亮度曲线预处理。如图7所示,亮度曲线的预处理流程从连续n帧历史帧提取目标图像序列,将提取的目标图像序列由彩色空间转换为灰度图,并计算每帧中目标区域的平均亮度值,获得一条随时间变化的表征目标在不同时间点的亮度特征亮度时序曲线。
受到环境光变化、临近车辆车灯影响、目标车辆局部遮挡以及摄像头自动曝光机制等因素的影响,从前述步骤获取的亮度波形曲线中夹杂一定的噪声干扰。为减弱噪声对检测结果的影响,采用以下步骤对原始亮度曲线进行预处理:
(1)SG(Savitzky-Golay)滤波。采用Savitzky- Golay滤波器(Press等,2007)对亮度时序信号进行平滑处理。该方法基于局部多项式拟合原理,在滑动窗口内执行多项式拟合运算,实现信号的平滑。滑动窗口的宽度和多项式的阶数依据波形数据的长度进行合理设定。此方法能够在有效去除随机噪声的同时,较好地保留曲线的本征变化趋势和周期性特征。
(2)去趋势化。实际场景中,检测框内的亮度值可能存在整体缓慢变化的趋势,会掩盖危险报警闪光灯信号的特征,对后续处理造成干扰。因此通过最小二乘拟合法拟合出一条代表整体变化趋势的直线,并将其从原始亮度信号中剔除,实现对亮度信号的去趋势化。该操作有助于消除低频背景成分的影响,使频谱分析更聚焦于真实的周期信号成分。
(3)梯度计算。为了更清晰地捕捉亮度随时间的变化特性,对时序亮度信号进行梯度运算,获取各时间点亮度的瞬时变化率。该处理方式可更直观地揭示危险报警闪光灯的闪烁节奏,使分析重点转向亮度变化速率而非其绝对值,有助于提升对周期性成分的响应敏感度。
2)曲线频谱分析。图8为曲线频谱分析的整体流程。在获取预处理后的亮度时序信号后,采用Welch方法(Welch,1967)对该信号进行功率谱密度(power spectral density,PSD)估计。对频谱结果执行主频峰值的检测,提取出信号中具有代表性的主导频率,据此判断该主频是否落在车辆危险报警闪光灯的典型频率范围内。
主导峰的识别过程包括两个关键环节。第1步:峰值定位,采用局部极大值搜索的方法,从功率谱密度曲线中提取出所有可能的峰点;第2步:对这些候选峰值进行显著性判别与主频匹配分析。峰值的显著性通过计算其与周围谷值之间的幅度差进行衡量,即衡量该峰相较于周边信号的突出程度,定量计算方法为
P=x[i]-max(Bleft, Bright)
式中,
x[i]
表示第i个检测到的峰值,
Bleft
和
Bright
分别对应其左右两侧的最低谷值。若某一峰值的突出度超过功率谱密度背景平均值的两倍,则可视为通过显著性评估并视为主导峰。仅当主导峰的频率落在车辆危险报警闪光灯的特定频段内,才认定该车辆开启闪光灯。该频率判定机制能有效减低噪声干扰和环境光波动带来的影响,提升对闪光灯信号的识别精度,详见第3.2节的实验细节。
选取两个夜间场景视频开展暗光疑似车辆亮度曲线分析,如图9所示,视频1和视频2均识别到违停车辆,同时也存在错误的暗光疑似车辆,分别采用不同颜色框进行标注。图10为对应的识别目标的时序平均亮度图。可以看出,车灯亮度闪烁时具有周期规律的特征,而背景干扰如普通路灯的亮度在时间轴上变化不大,进行亮度时序曲线频谱分析,能有效区分暗光疑似车辆与背景干扰。
鉴于目前尚无面向高速公路违停检测任务的公开标准数据集,结合实际高速公路监控应用的需求,本文首次构建了3个针对性强、标注精细的高质量专用数据集:高速公路车辆目标检测数据集、车辆跟踪数据集和车辆违停识别视频数据集,涵盖车辆目标检测、跟踪以及违停识别等任务。数据均来源于真实高速公路监控视频片段,充分覆盖不同时间段(白天/夜间)、多种道路类型(主线/隧道)及复杂气象条件(雨雾天气),视频帧率(frame per second,fps)为25帧/s,分辨率为1 920 × 1 080像素或1 280 × 720像素。
1)车辆目标检测数据集。包含8 532幅图像,涵盖白天、夜间及雨天等多样化场景,按7∶3比例划分为训练集与测试集。夜间或低照度场景的车辆整体轮廓模糊,画面易表现为模糊光斑,单帧下难以准确识别,本文将该类目标定义为“暗光疑似车辆”并划分为一类,旨在降低夜间漏检率,同时结合时序信号分析,提升夜间识别的可靠性。表1列出了该数据集各类别目标框的数量分布,包括小型车辆、大型货车、大巴车辆以及暗光疑似车辆4类目标。图11展示了4类目标的示例图像,用红色方框标注。
2)高速公路车辆跟踪数据集。由训练子集与测试子集构成,视频片段的平均时长在30 s以上。通过以5帧/s的频率抽取图像帧,对车辆目标进行帧号标注、身份ID分配以及边界框位置的人工注释。表2列出了训练集与测试集分别包含的视频片段数量、图像帧总数以及对应的车辆轨迹数量。
3)高速公路车辆违停识别数据集。共收录了77段时长超过1 min的视频,涵盖98辆发生违停行为的车辆,且每辆车的违停持续时间均超过10 s。为评估所提出算法在多种光照条件和天气环境下的适应能力,特别是在夜间环境中仍具备良好检测性能和较低误报率,数据集特意选取了来自不同路段与时段的视频资料。该数据集具体划分为夜间、白天和隧道3类典型场景。各场景中视频片段数量及对应的违停车辆统计信息详见表3。为进一步检验算法在复杂天气条件下的鲁棒性,每类场景中还包含了如雨天、浓雾等恶劣气象条件下的样本片段。图12展示了部分典型场景的示例图像。
1)硬件配置。系统为Ubuntu20.04;CPU配置2个英特尔Xeon Silver 4210;内存共为32 GB;并配置一张RTX 5000 GPU,显存为16 GB GDDR6。
2)评估指标。为了全面评估高速公路车辆违停识别算法的性能,采用了准确率(precision,P)、召回率(recall,R)、F1分数、使用显存以及fps指标,对车辆违停判别模块的有效性进行量化分析。若系统在车辆违停期间成功识别该目标,则视为一次正确检测(true positive,TP),否则记为漏检(false negative,FN)。此外,为验证所引入的双重注意力特征融合机制对车辆跟踪结果的增强效果,进一步采用多目标跟踪准确率(multiple object tracking accuracy,MOTA)、身份F1分数(ID F1 score,IDF1)以及高阶跟踪精度(higher order tracking accuracy,HOTA)(Luiten等,2021)3项主流指标,对车辆跟踪子模块的性能进行系统性评估。同时引入帧率(帧/s)指标以衡量跟踪算法速度。
3)实验细节。根据实际场景的应用需求,车辆判定为违停的最小静止时间设定为D = 10 s。考虑到车辆从运动状态过渡到完全静止的过程,累计静止帧的阈值β设定为0.8 × D。在帧间自适应位移判定模块中,参数设定为
wup
= 0.25,
wdown
= 0.75,
τα
=
τN
= 1。平滑因子
γ
由式(5)定义给出,其中平滑参数
λ
= 0.01,以避免分母接近零而导致的数值不稳定问题。对于多项式阶数k,本文对比了不同k值的车辆检测算法效果,如表4所示。
从表4可以看出,当k = 3时,检测算法效果最好,后续研究选择k = 3。在基于时序亮度信号进行分析的验证阶段,借助Python中scipy.signal库实现各关键处理步骤。Savitzky-Golay滤波使用savgol_filter函数完成,其中sample_rate为视频帧率,窗口长度window_length设定为sample_rate/4;趋势项去除使用detrend函数实现;welch功率方法使用welch函数实现,窗口大小window设定为时序信号长度的1/4;局部峰值的提取通过find_peaks函数完成,其返回的显著性数值用于判断峰值的重要程度。依据国家标准,汽车电子闪光器的正常工作频率为每分钟60~120次,折算为1~2 Hz。将车辆危险报警闪光灯的主频判定区间设定为[0.5,2] Hz,确保检测系统对合法频率范围内的闪烁模式具备良好的识别能力。
表5展示了所提出的车辆违停识别算法在自构建的高速公路违停识别数据集上的实验结果。在夜间开启危险报警闪光灯违停的测试场景中,共有25段视频在目标成功被检测的前提下均实现了准确识别,另外2段因目标车辆未能被检测出,系统无法完成违停判断。
部分典型的检测成功与失败案例如图13所示。其中,第1行为检测成功的例子。第1列显示车辆停在应急车道,旁边车道持续有车辆经过;第2列显示夜间车辆停在应急车道开启双闪,缺乏清晰的车辆轮廓信息;第3列显示车辆多次在应急车道缓行和停止,均被检测到。第2行为检测失败的例子。第1列图像显示道路远端的两车相撞,其中1辆无法识别;第2列显示多辆违停车辆存在遮挡状态,部分违停车辆无法识别;第3列显示隧道远端持续缓行的货车,自适应帧间位移无法判别其处于运动的状态导致误报。其中,漏检的主要原因包括车辆因遮挡或处于画面远端尺寸过小,导致目标检测失败;而误检情况则多由车辆缓慢移动而非真正静止引起,系统将其判断为违停。这些失败的样例在后续对比实验所采用的4种方法中同样未能正确识别,体现出此类边界场景的挑战性。
该实验结果表明,本文算法在召回率与准确率之间实现了较理想的权衡,在保持较高检出能力的同时,有效降低了误判的发生概率,展现出良好的整体检测性能。
为评估所提出的基于双重注意力特征融合机制的目标跟踪方法的实际效果,本文选取了现有多目标跟踪领域中6种具有代表性的算法作为对比基准,分别为BotSORT(bag of tricks SORT)(Aharon等,2022)、Bytetrack(Zhang等,2022)、StrongSORT(stronger,faster and more robust SORT)(Du等,2023)、Deep OC-SORT(deep observation-centric SORT)(Maggiolino等,2023)、MambaTrack(Xiao等,2024)和BoostTrack++(Stanojević和Todorović,2024)。
在实验过程中,首先在构建的高速公路车辆检测数据集上对目标检测模型进行训练,用于提取外观特征的子网络同样也在高速公路车辆多目标跟踪训练集上完成训练。最终,将各方法在高速公路车辆多目标跟踪测试集上进行实验对比,其性能指标结果如表6所示。从实验结果可以看出,提出的算法在主要评估指标上均展现出较强的性能。在HOTA指标上,本文方法达到了66.3%的得分,检测效果和运算速度对比BotSORT具有明显的提升,虽然BoostTrack++在HOTA、MOTA和IDF1上表现最优,但是其运算速度明显比提出的算法慢,进一步验证了本文方法在同时考虑检测精度与数据关联准确性方面具备更强的综合优势。这一结果表明,提出的算法在单帧检测上表现优秀,增强了在目标跟踪过程中保持身份相对稳定的能力,降低了身份切换频率与轨迹碎片的数量,提升了整个跟踪流程的连贯性。
为验证所提出的高速公路车辆违停识别方法的有效性,本节将其与4种现有的违停识别技术进行对比实验,比较整体算法流程的性能表现。具体而言,方法1(梁睿琳 等,2022)采用YOLOv4对车辆进行检测,借助透视关系建模将远处与近处的车辆目标统一到相同的尺度体系,构建时空信息矩阵进行违停识别;方法2(Hamzah等,2022)依托Inception V2(Ioffe和Szegedy,2015)与Faster R-CNN(Ren等,2017)框架,将车辆违停识别问题转化为一个图像分类任务;方法3(陈伟 等,2024)引入膨胀空间金字塔池化(dilated spatial pyramid pooling,DSPP)和带分支注意力机制的膨胀空间金字塔池化(dilated spatial pyramid pooling with branch attention,DSPPA)模块改进YOLOv8网络的特征金字塔池化层,并设计基于全局匹配的停车位分配策略进行违规停车的检测;方法4(Kuo和Lin,2024)以YOLOP作为网络架构框架,通过整合YOLOv7中的E-ELAN模块和RepConvN等模块,优化模型架构,降低参数和浮点运算。各项对比实验的评估结果如表7所示。可以看出,提出的检测方法在准确率、召回率以及F1分数这3项核心评价指标上均取得了最优表现,整体优于所选的4种对比方案。其中,方法1基于透视模型配合YOLOv4的策略,存在较严重的漏检现象,误报率偏高;使用Faster R-CNN结合InceptionV2的方法2略有提升,但整体精度仍然不高;方法3采用YOLOv8结合全局匹配策略的算法,召回率和处理速度比方法2效果好,但准确率略低;方法4的检测效果在4个比较算法中的检测效果最好,但夜间场景下不佳,存在较严重的漏检误检。
相比之下,本文方法在检测完整性与准确性之间实现更理想的平衡,准确率提升至94.79%,召回率达到92.86%,F1分数为0.94,表明本文方法在有效降低漏检数量的同时,并未引入额外的误报,并且算法速度上仅低于方法3,验证了算法在实际应用中的可靠性与优越性。
同时,从表7可以看出,本文算法在子场景性能一致性上表现突出,夜间、白天、隧道子场景的F1 分数分别达0.91、0.95、0.92,最大值与最小值差值仅0.04,远优于方法1(差值0.19)、方法2(差值0.18)、方法3(差值0.15)和方法4(差值0.16),稳定性源于次级验证模块的作用。该模块通过分析亮度时序信号的周期性与稳定性,有效抑制夜间路灯反光、隧道照明光斑等场景干扰,降低夜间、隧道子场景的误判率,说明算法在数据分布差异较大的环境中能够保持稳定性能,体现了算法的跨场景适应性。
为验证双重注意力特征融合模块中空间注意力子模块、通道注意力子模块的独立贡献及两者组合的协同效应,本研究设计了该模块的消融实验,分析各模块对高速公路车辆跟踪性能的影响。
研究选用YOLOv8作为主干检测网络,搭配BotSORT-ReID实现车辆跟踪与身份关联。组别1作为无注意力机制的基线模型;组别2(仅SAM)仅在 YOLOv8的多尺度特征层(1/8、1/16、1/32)嵌入空间注意力子模块(SAM),不引入通道注意力;组别3(仅 SEBlocks)则在 YOLOv8-s 的特征融合层嵌入通道注意力子模块,不引入空间注意力;组别4(DAFM)为本文完整的设计。
实验结果如表8所示。可以看出,以基线(组别1)为参照,单独加入SAM(组别2)时,MOTA 提升1.1%、IDF1提升4.5%,这是由于SAM子模块抑制背景与车辆遮挡,提升了定位准确性;单独加入 SEBlocks(组别3),由于可筛选车辆关键特征通道,减少远端小目标漏检,指标有一定的提升。SAM与SEBlocks组合的完整DAFM(组别4)所有指标最优,表明二者形成互补优化,有效应对了高速公路多尺度车辆、密集遮挡等场景痛点。
为全面评估各功能模块在提升高速公路车辆违停识别准确性方面的实际价值,开展一系列消融实验,围绕提出的3项关键创新模块展开验证,包括:基于双重注意力特征融合的表观特征提取模块、自适应的帧间位移判定机制,以及利用时序信号分析进行的次级判别模块。实验中所采用的基线方案以YOLOv8作为车辆检测器,借助BotSORT进行多目标跟踪,当相邻两帧的交并比(intersection over union,IoU)大于0.3时,判断车辆未发生位移,若该状态持续超过10 s,则认定为违停。在此基础上,依次引入上述3个功能模块。方案1在基线方法的基础上进行了两项改进:在目标检测类别中新增“暗光疑似车辆”类别,适应低照度场景下的检测需求;同时引入双重注意力特征融合的表观特征提取模块,验证增强目标在轨迹关联过程中的特征表达能力。方案2在方案1的基础上,将IoU阈值的判定标准替换为自适应帧间位移策略,评估该机制在应对检测框轻微抖动和提升判定稳定性方面的效果。方案3进一步集成基于时序信号分析的危险报警闪光灯检测模块,形成完整的检测流程。
表9展示了各项实验方案的具体结果。基线方法在违停识别任务中的表现一般,召回率仅为75.51%。方案1在保留基线架构的基础上引入表观特征提取模块,召回率提升了5.1%,主要得益于跟踪阶段车辆身份保持能力的增强,使得基线中因ID混淆导致的漏检目标能够被重新捕捉。但由于增加新类别,使得部分类似反光标识的非车辆目标被误判,拉低准确率。方案2在此基础上引入自适应帧间位移判断机制,使违停识别对框体抖动等干扰更加鲁棒。具体表现为准确率提高至80.9%,F1值提升至0.86,漏检现象显著减少。方案3集成危险报警闪光灯检测模块,有效解决夜间场景下误检频发的问题,3个指标均为最优,F1分数达到0.94。数据说明暗光疑似类别与双闪检测模块形成良好的协同效应,有效提升夜间场景的检测效果,验证了所提出方法的可靠性与优越性。
图14展示了车辆之间发生遮挡时的典型情形。其中,第1行是视频第190帧画面,应急车道上违停的轿车和正常通行的车辆均被检测并被赋予身份编号。第2行对应第200帧,一辆行驶中的货车遮挡了违停车辆。此时基线方法的跟踪框出现误关联,主要原因是夜间光线较差,车辆的外观特征不够显著。方案1引入表观特征提取模块有效缓解了该问题,避免了ID匹配错误引发的漏检。
图15的白天场景显示画面远端存在正在行驶的货车,基线算法和方案1使用IoU判定策略,导致产生误检。方案2引入自适应帧间位移判别方法,能够有效地适应距离产生的透视畸变,成功识别了该货车为低速移动状态,并以蓝色框进行标注。
图16展示了引入的基于时序信号频谱分析的二次验证模块的实际效果。其中,第1行和第2行分别显示强烈反光的路牌和夜间远处路口区域持续出现光斑,均误识别为暗光疑似车辆类别,而方案1和方案2均错误判别违停。方案3经过时序信号频谱分析的验证后准确排除该误检。
针对高速公路违停检测的身份混淆、透视变形及夜间场景鲁棒性不足等问题,本文提出了融合双重注意力特征和时空特征的违停识别算法。首先设计了双重注意力特征融合模块以强化关键特征表达,然后通过自适应帧间位移算法处理透视变形和框抖动问题,并提出基于时序信号分析的闪光灯检测算法,提升夜间判定效果。针对公开数据资源的不足,构建了具有代表性涵盖多场景的数据集,为算法评估提供可靠支撑,提升了研究的可复现性与实用性。通过该高速公路违停数据集,评估了所提出的算法性能。实验结果表明,该算法在不同时间段和天气条件下具有良好的鲁棒性,准确率和召回率表现优异。未来将扩展数据集规模,增加跨区域与跨时段样本,提升算法的泛化能力和适应性。
  • 广东省省级科技计划项目(2023A1111120018)
参考文献 引证文献
排序方式:
Aharon N,Orfaig R and Bobrovsky B Z. 2022. BoT-SORT: robust associations multi-pedestrian tracking [EB/OL]. [2025-07-15]. http://arxiv.org/pdf/2206.14651v1.pdf
Chen W,Wang X L,Zhang Y W,An G C and Jiang B. 2024. Vehicle illegal parking detection in highway service areas based on improved YOLOv8. Computer Engineering,50(4): 11-19
陈伟,王晓龙,张晏玮,安国成,江波. 2024. 基于改进YOLOv8的高速公路服务区车辆违停检测. 计算机工程,50(4): 11-19[DOI: 10.19678/j.issn.1000-3428.0068901]
Du Y H,Zhao Z C,Song Y,Zhao Y Y,Su F,Gong T,et al. 2023. StrongSORT: make DeepSORT great again. IEEE Transactions on Multimedia,25: 8725-8737 [DOI: 10.1109/TMM.2023.3240881]
Hamzah R A,Setianingsih C,Nugrahaeni R A,Hanafia S R and Fuadi F. 2022. Parking violation detection on the roadside of toll roads with intelligent transportation system using faster R-CNN algorithm//Proceedings of the 6th International Conference on Informatics and Computational Sciences. Semarang,Indonesia: IEEE: 169-174 [DOI: 10.1109/ICICoS56336.2022.9930590]
Hassan W,Birch P,Young R and Chatwin C. 2012. Real-time occlusion tolerant detection of illegally parked vehicles. International Journal of Control,Automation and Systems,10(5): 972-981 [DOI: 10.1007/s12555-012-0514-2]
Hu J,Shen L,Albanie S,Sun G and Wu E H. 2020. Squeeze-and-excitation networks. IEEE Transactions on Pattern Analysis and Machine Intelligence,42(8): 2011-2023 [DOI: 10.1109/TPAMI.2019.2913372]
Ioffe S and Szegedy C. 2015. Batch normalization: accelerating deep network training by reducing internal covariate shift//Proceedings of the 32nd International Conference on Machine Learning. Lille,France: JMLR.org: 448-456
Ketcham M,Getkhaw E,Piyaneeranart M,Ganokratanaa T and Yimyam W. 2019. Recognizing the illegal parking patterns of cars on the road in front of the bus stop using the support vector machine//Proceedings of the 15th International Conference on Signal-Image Technology and Internet-Based Systems. Sorrento,Italy: IEEE: 538-542 [DOI: 10.1109/SITIS.2019.00091]
Kuo L C and Lin H Y. 2024. Illegal parking detection based on multi-task driving perception//Proceedings of 2024 IEEE Intelligent Vehicles Symposium. Jeju Island,Korea (South): IEEE: 1865-1870 [DOI: 10.1109/iv55156.2024.10588798]
Liang R L,Wang R and Guo Y. 2022. Highway abnormal parking detection based on video spatio-temporal relationship. Application Research of Computers,39(6): 1916-1920
梁睿琳,王锐,郭迎. 2022. 基于视频时空关系的高速公路异常停车检测. 计算机应用研究,39(6): 1916-1920[DOI: 10.19734/j.issn.1001-3695.2021.09.0617]
Luiten J,Ošep A,Dendorfer P,Torr P,Geiger A,Leal-Taixé L,et al. 2021. HOTA: a higher order metric for evaluating multi-object tracking. International Journal of Computer Vision,129(2): 548-578 [DOI: 10.1007/s11263-020-01375-2]
Maggiolino G,Ahmad A,Cao J K and Kitani K. 2023. Deep OC-SORT: multi-pedestrian tracking by adaptive re-identification//Proceedings of 2023 IEEE International Conference on Image Processing. Kuala Lumpur,Malaysia: IEEE: 3025-3029 [DOI: 10.1109/ICIP49359.2023.10222576]
Matsuda A,Matsui T,Matsuda Y,Suwa H and Yasumoto K. 2021. A system for real-time on-street parking detection and visualization on an edge device//Proceedings of 2021 IEEE International Conference on Pervasive Computing and Communications Workshops and Other Affiliated Events. Kassel,Germany: IEEE: 227-232 [DOI: 10.1109/PerComWorkshops51409.2021.9431076]
Park H,Kim K,Jeong I,Jung J and Cho J. 2025. Special vehicle classification algorithm-based system for dedicated parking zone violation detection in South Korea. IEEE Access,13: 7883-7901 [DOI: 10.1109/ACCESS.2025.3526862]
Peng X G,Song R Z H,Cao Q,Li Y,Cui D S,Jia X F,et al. 2022. Real-time illegal parking detection algorithm in urban environments. IEEE Transactions on Intelligent Transportation Systems,23(11): 20572-20587 [DOI: 10.1109/TITS.2022.3180225]
Press W H,Teukolsky S A,Vetterling W T and Flannery B P. 2007. Numerical Recipes: The Art of Scientific Computing. 3rd ed. Cambridge,UK: Cambridge University Press [DOI: 10.1007/BF01321860]
Ren S Q,He K M,Girshick R and Sun J. 2017. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence,39(6): 1137-1149 [DOI: 10.1109/TPAMI.2016.2577031]
Stanojević V and Todorović B. 2024. BoostTrack++: using tracklet information to detect more objects in multiple object tracking [EB/OL]. [2025-07-15]. http://arxiv.org/pdf/2408.13003v1.pdf
Wahyono,Filonenko A and Jo K H. 2016. Detecting illegally parked vehicle based on cumulative dual foreground difference//Proceedings of the 14th IEEE International Conference on Industrial Informatics. Poitiers,France: IEEE: 772-775 [DOI: 10.1109/INDIN.2016.7819263]
Wang H,Liu J,Zhao J,Zhang J and Zhao D. 2025. Precision and speed: LSOD-YOLO for lightweight small object detection. Expert Systems with Applications,269: #126440 [DOI: 10.1016/j.eswa.2025.126440]
Wang Q,Xue X Y,Min W D,Wang S,Gai D and Han Q. 2024. Cross-domain joint learning and shared subspace metric for vehicle re-identification. Journal of Image and Graphics,29(5): 1364-1380
汪琦,雪心远,闵卫东,汪晟,盖迪,韩清. 2024. 跨域联合学习与共享子空间度量的车辆重识别. 中国图象图形学报,29(5): 1364-1380 [DOI: 10.11834/jig.230405]
Welch P. 1967. The use of fast Fourier transform for the estimation of power spectra: a method based on time averaging over short,modified periodograms. IEEE Transactions on Audio and Electroacoustics,15(2): 70-73 [DOI: 10.1109/TAU.1967.1161901]
Woo S,Park J,Lee J Y and Kweon I S. 2018. CBAM: convolutional block attention module//Proceedings of the 15th European Conference on Computer Vision. Munich,Germany: Springer: 3-19 [DOI: 10.1007/978-3-030-01234-2_1]
Xiao C C,Cao Q,Luo Z G and Lan L. 2024. MambaTrack: a simple baseline for multiple object tracking with state space model//Proceedings of the 32nd ACM International Conference on Multimedia. Melbourne,Australia: ACM: 4082-4091 [DOI: 10.1145/3664647.3680944]
Xie X M,Wang C Y,Chen S,Shi G M and Zhao Z F. 2017. Real-time illegal parking detection system based on deep learning//Proceedings of 2017 International Conference on Deep Learning Technologies. Chengdu,China: ACM: 23-27 [DOI: 10.1145/3094243.3094261]
Xu H,Dong S H,Zhang J W and Zheng Y H. 2025. Context-aware attention fused Transformer tracking. Journal of Image and Graphics,30(1): 212-224
徐晗,董仕豪,张家伟,郑钰辉. 2025. 融合上下文感知注意力的Transformer目标跟踪方法. 中国图象图形学报,30(1): 212-224 [DOI: 10.11834/jig.240084]
Yu Y,Chen W X and Chen F X. 2023. RIC-NVNet: night-time vehicle enhancement network for vehicle model recognition. Journal of Image and Graphics,28(7): 2054-2067
余烨,陈维笑,陈凤欣. 2023. 面向车型识别的夜间车辆图像增强网络RIC-NVNet. 中国图象图形学报,28(7): 2054-2067[DOI: 10.11834/jig.220122]
Zeng F G,Dong B,Zhang Y,Wang T C,Zhang X Y and Wei Y C. 2022. MOTR: end-to-end multiple-object tracking with Transformer//Proceedings of the 17th European Conference on Computer Vision. Tel Aviv,Israel: Springer: 659-675 [DOI: 10.1007/978-3-031-19812-0_38]
Zhang Y F,Sun P Z,Jiang Y,Yu D D,Weng F C,Yuan Z H,et al. 2022. ByteTrack: multi-object tracking by associating every detection box//Proceedings of the 17th European Conference on Computer Vision. Tel Aviv,Israel: Springer: 1-21 [DOI: 10.1007/978-3-031-20047-2_1]
2026年第31卷第7期
PDF下载
23
10
引用本文
BibTeX
文章信息
doi: 10.11834/jig.250354
  • 接收时间:2025-07-29
  • 首发时间:2026-10-30
  • 出版时间:2026-07-16
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2025-07-29
  • 修回日期:2025-12-06
基金
Guangdong Provincial Science and Technology Project(2023A1111120018)
广东省省级科技计划项目(2023A1111120018)
作者信息
    1 华南理工大学计算机科学与工程学院,广州 510006
    2 广州国交润万交通信息有限公司,广州 510665

通讯作者:

参考文献
分享链接
https://castjournals.cast.org.cn/joweb/zgtxtxxb/CN/10.11834/jig.250354
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
科
Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)
属
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