测绘通报 ›› 2024, Vol. 0 ›› Issue (5): 147-150.doi: 10.13474/j.cnki.11-2246.2024.0526

• 技术交流 • 上一篇    

基于轻量化车载设备的道路病害检测方法

姚楚羡1, 蔡皓楠1, 张远波1, 唐可懿1, 詹璐1, 周宝定1,2   

  1. 1. 深圳大学土木与交通工程学院, 广东 深圳 518060;
    2. 深圳大学城市智慧交通与安全运维研究院, 广东 深圳 518060
  • 收稿日期:2023-11-01 发布日期:2024-06-12
  • 通讯作者: 周宝定。E-mail:bdzhou@szu.edu.cn
  • 作者简介:姚楚羡(2001—),女,硕士生,研究方向为智能交通。E-mail:yaochuxian2019@email.szu.edu.cn
  • 基金资助:
    广东省科技创新战略专项资金(大学生科技创新培育)(pdjh2022a0435)

Road damage detection method based on lightweight vehicle equipment

YAO Chuxian1, CAI Haonan1, ZHANG Yuanbo1, TANG Keyi1, ZHAN Lu1, ZHOU Baoding1,2   

  1. 1. College of Civil and Transportation Engineering, Shenzhen University, Shenzhen 518060, China;
    2. Institute of Urban Smart Transportation & Safety Maintenance, Shenzhen University, Shenzhen 518060, China
  • Received:2023-11-01 Published:2024-06-12

摘要: 针对传统道路检测方法成本高、检测周期长,无法满足城市道路大规模、短周期检测需求的问题,本文设计了一种轻量化车载道路病害数据获取设备,并提出了基于轻量化车载设备的道路病害检测方法。该设备质量小、成本低,可简易、快速地安装在小汽车、公共汽车等城市常见车辆上,能够同步收集惯性数据、图像、GPS定位信息等数据。基于轻量化车载设备在城市道路上采集道路图片数据并构建数据集,然后建立深度学习模型对其进行训练和评估,检测和识别道路病害,正确率达82.54%,能够满足城市道路日常巡检的要求。

关键词: 道路工程, 道路病害检测, 轻量化车载设备, 深度学习, 目标检测

Abstract: In response to the high cost and long detection cycle of traditional road detection methods, which can not meet the needs of large-scale and short-term detection of urban roads, this paper designs a lightweight vehicle mounted road damage data acquisition device and proposes a road damage detection method based on lightweight vehicle mounted devices. This device has the characteristics of small mass and low cost, and can be easily and quickly installed on common urban vehicles such as cars and buses, and synchronously collect inertial data, images, GPS positioning information, and other data. The road damage detection method proposed in this article is based on lightweight vehicle mounted devices, which collect road image data on urban roads to construct a dataset, establish a deep learning model for training and evaluation, and detect and recognize road damage. The accuracy rate is 82.54%, which can meet the requirements of daily inspection of urban roads.

Key words: road engineering, road damage detection, lightweight on-board equipment, deep learning, target detection

中图分类号: