基于YOLOV5网络模型的市政道路检测识别

Municipal Road Detection and Recognition Based on YOLOV5Network Model

根据十四五规划中提出的加快智能化城市建设服务体系,市政道路是智慧城市建设的重要内容,道路损坏检测则是保证市政道路良好管理的重要手段。本文利用YOLOV5算法经过数据集制作、实验设定、模型训练、问题识别等过程对呼和浩特市市政道路损害情况识别,识别后通过系统及时告知维修人员检修,大大节省了人力和时间成本。对数据集中2000多张市政道路图像进行标注及分类及模型训练,实验表明,本文的模型mAP达到了0.9以上,该方法可以提高市政道路图像的检测精度。

Detection and identification of municipal roads based on yolov5network model according to the service system of accelerating intelligent urban construction proposed in the 14th five year plan,municipal roads are an important part of smart city construction,and road damage detection is an important means to ensure good management of municipal roads.In this paper,yolov5algorithm is used to identify the damage of municipal roads in Hohhot through the process of data set production,experimental setting,model training and problem identification.After identification,the maintenance personnel are informed of maintenance in time through the system,which greatly saves manpower and time costs.More than 2000municipal road images in the data set are labeled,classified and trained.The experiments show that in the final model evaluation and test,the real positive samples in the precision sample are predicted to be more than 0.9,and the positive samples in the recall rate of more than 0.9are predicted correctly.