摘要
《郑州市非机动车管理办法》中规定,电动车骑乘人员不戴头盔,处以警告或者罚款。但由于河南省郑州市的电动车数量庞大,交警部门很难确保该办法能够顺利实施。针对该问题,基于特征金字塔网络对传统单镜头多盒检测器(Single Shot Multibox Detector,SSD)模型进行改进,以改进模型为基础搭建了头盔佩戴检测模型,通过该模型来帮助交警部门提升工作效率。最终测试结果显示,该模型与传统SSD模型及YOLOV5模型相比,其识别准确率分别提升9.84%和2.09%。
The"Zhengzhou Non-Motor vehicle Management Measures"has provisions that electric bicycle riders not wearing helmets would be imposed warnings or fines.However,due to the large number of electric bicycles,it’s difficult for the police to ensure the implementation of the policy.In order to solve this problem,this paper improves the Single Shot Multibox Detector(SSD)model based on the feature pyramid network,and builds a helmet wearing detection model based on the improved model.Through it to help the police to improve work efficiency.The final test results show that compared with the traditional SSD model and YOLOV5 model,the recognition accuracy of this model is 9.84%and2.09%relatively improved.
作者
卢云聪
LU Yuncong(Intelligent Police Research Center,Railway Police College,Zhengzhou Henan 450000,China)
出处
《信息与电脑》
2022年第16期14-16,共3页
Information & Computer
基金
2022年中央高校基本科研业务项目(项目编号:2022TJJBKY025)。
