基于YOLOv5的车流量检测研究  

Research on traffic flow detection based on YOLOv5

作　　者：吴文静 章培军[1] Wu Wenjing;Zhang Peijun(School of Computer Science,Xijing University,Xi’an 710123,China)

机构地区：[1]西京学院计算机学院,西安710123

出　　处：《现代计算机》2024年第12期24-27,共4页Modern Computer

摘　　要：对基于卷积神经网络的目标检测以及流量检测原理进行研究,选择在实时性、精确性等方面具有良好表现的YOLOv5作为目标检测模型。然后采用Labelme对采集到的交通视频数据进行标注,以PyCharm作为集成开发环境进行编程,实现了车辆目标检测与流量检测两个主要功能。测试运行结果表明,不管在白天还是夜晚,在各种干扰条件下,基于YOLOv5的视频车流量检测软件都可以准确检测出当前车流量,具有较强的鲁棒性、实时性和精确性,可满足实际应用的需求。The principle of target detection and traffic detection based on convolutional neural network is studied,and YOLOv5,which has good performance in real-time and accuracy,is selected as the target detection model.Then label the collected traffic video data with Labelme,and program with PyCharm Community Edition 2020 as the experimental environment to realize the two main functions of vehicle target detection and flow detection.The test results show that the video traffic flow detection software based on YOLOv5 can accurately detect the current traffic flow no matter in the daytime or at night under various interference conditions,and has strong robustness,real-time and accuracy,which can meet the needs of practical application.

关 键 词：交通视频 车辆检测 车流量统计 深度学习 YOLOv5 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程] U495[交通运输工程—交通运输规划与管理]