基于改进Faster R-CNN的道路坑洞检测方法  

Road Pothole Detection Method Based on Improved Faster R-CNN

作　　者：黄尚安 阮太元[1] 周昊[1] 黄俊 蓝伟劲 Huang Shang'an;Ruan Taiyuan;Zhou Hao;Huang Jun;Lan Weijin(Jiangmen Polytechnic,Jiangmen,China)

机构地区：[1]江门职业技术学院,广东江门

出　　处：《科学技术创新》2023年第9期112-115,共4页Scientific and Technological Innovation

基　　金：2021年广东省科技创新战略专项资金(大学生科技创新培育)——《基于无人机+5G+Al的高速公路突发事件监测预警系统设计》;2021年广东省普通高校青年创新人才类项目——《基于无人机的高速公路智能预警系统设计研究》。

摘　　要：道路坑洞会造成大量的交通事故及人民群众生命财产损失。道路坑洞的修补首先要进行检测,而完全依赖人工则会效率低成本高,提出一种基于改进Faster R-CNN的道路坑洞检测方法,在公共数据集[1]上的精度达到了83.6%。Roads are one of the basic constructions in China's social economy and the most basic transportation platform for various goods.However,with the increase of service life,vehicle overloading,or engineering quality,various potholes may appear.These road potholes will cause a large number of traffic accidents and loss of people's lives and property.The repair of road potholes should first be detected,and it will be low efficiency and high cost to rely entirely on manual work.A road potholes detection method based on improved Faster R-CNN is proposed.After the modification of anchor points and backbone networks,the accuracy on the public data set[1]has reached 83.6%.

关 键 词：道路 Faster R-CNN 坑洞检测 

分 类 号：U418.6[交通运输工程—道路与铁道工程] TP391[自动化与计算机技术—计算机应用技术]