摘要
文章提出了一种车辆检测网络模型,旨在更好地检测小型车辆目标,提高目标检测的准确性,从而实现更具鲁棒性的异常检测。同时,文章还探讨了一种基于检测框匹配的异常停车检测方法,以期提升异常检测的准确性。为了优化异常检测方法,本研究引入车辆检测网络模型于检测车辆目标,并结合目标检测和目标跟踪方法生成道路掩码。通过综合应用,进一步完善异常检测器,实现对异常事件开始时间和结束时间的准确获取。
This article proposes a vehicle detection network model aiming at better detecting small vehicle targets,improving the accuracy of object detection,and achieving more robust anomaly detection.At the same time,the paper also explores an anomaly parking detection method based on detection box matching to improve the accuracy of anomaly detection.In order to optimize anomaly detection methods,a vehicle detection network model was introduced in the study to detect vehicle targets,and road masks were generated by combining target detection and target tracking methods.Through the comprehensive application of these methods,the anomaly detector has been further improved,achieving accurate acquisition of the start time and end time of abnormal events.
作者
李莹莹
李海芳
宋瑞霞
刘战东
李克
丁男
LI Ying-ying;LI Hai-fang;SONG Rui-xia;LIU Zhan-dong;LI Ke;Ding Nan(College of Computer Science and Technology,Xinjiang Normal University,Urumqi,Xinjiang,830054,China;Library,Xinjiang Normal University,Urumqi,Xinjiang,830017,China)
出处
《新疆师范大学学报（自然科学版）》
2025年第2期8-19,共12页
Journal of Xinjiang Normal University(Natural Sciences Edition)
基金
国家自然科学基金项目(62162061,62262066)
新疆维吾尔自治区自然科学基金项目(2021D01E20)
新疆维吾尔自治区重点研发计划项目(2022B01007-1,2022B01007-3)
新疆师范大学博士启动基金项目(XJNUBS2115)
新疆师范大学青年拔尖人才项目(XJNUQB2022-21)。
关键词
计算机视觉
目标检测
异常检测
背景建模
道路掩码
Computer vision
Object detection
Abnormal detection
Background modeling
Road mask
