智能电子警察系统设计与实现

Design and Implementation of Intelligent Electronic Police System

摘要: 随着社会经济高速发展，交通在社会活动中的地位日益显著。机动车的数量呈现不断增加的趋势，传统的电子警察系统已经无法满足现代化交通运输需求，车辆违章行为愈发频繁。为了降低车辆违章行为的发生率，提高电子警察系统的预警能力，本文基于改进的YOLOv5s+DeepSort神经网络建立车辆检测与跟踪的预测模型，并结合CDNet神经网络进行斑马线检测以及Ultra Fast Structure-aware Deep Lane Detection神经网络进行车道线检测，设计并实现了一款智能电子警察系统。系统根据检测目标实现了超速检测、压线检测、闯红灯检测、逆行检测、礼让行人检测、违章掉头检测、不按车道行驶检测、行人闯红灯检测、违停检测9种违章行为检测。该系统采用Vue2.0渐进式框架、Flask框架、SpringBoot等前后端开发技术，设计实现视频图像分析处理、预警、违章管理等模块，实时检测违章行为，并提供预警机制及时提醒用户违章车辆的出现。同时，系统提供流量统计模块，收集车流量信息，分析道路拥堵情况，助力交警智能决策。

Abstract: With the rapid development of the social economy, transportation has become increasingly prominent in social activities. The number of motor vehicles is increasing. The traditional electronic police system has been unable to meet the needs of modern transportation, and vehicle violations are increasingly frequent. In order to reduce the incidence of vehicle violations and improve the early warning ability of electronic police systems, this paper designs and implements an intelligent electronic police system based on the improved YOLOv5s + DeepSort neural network to establish the prediction model of vehicle detection and tracking, and combines CDNet neural network for zebra line detection and Ultra Fast Structure-aware Deep Lane Detection neural network for lane line detection. According to the detection targets, the system realizes nine kinds of illegal behavior detection, including overspeed detection, line detection, red light detection, retrograde detection, courtesy pedestrian detection, illegal turn detection, non-lane driving detection, pedestrian red-light detection, and illegal stop detection. The system adopts Vue2.0 progressive framework, Flask framework, SpringBoot and other front-end and back-end development technologies to design and implement video image analysis and processing, early warning, violation management and other modules, detect violations in real time, and provide an early warning mechanism to remind users of the appearance of illegal vehicles in time. Meanwhile, the system provides a flow statistic module to collect traffic flow information, analyze road congestion, and help traffic police intelligent decision-making.