基于无人机视频的公路桥梁边坡异常检测

Abnormal Detection of Highway Bridge and Slope Based on UAV Video

对无人机巡查相应的软硬件系统进行开发,实现了对公路桥梁边坡中常见的边坡坡面冲刷,桥梁地基冲刷以及桥下空间违规停放车辆、违章建筑、堆积物等异常隐患的高精度自动化检测。首先搭建了高精度无人机控制系统、地面工作站和数据管理中心,用于自动化采集桥梁边坡影像;然后构建桥梁边坡常见异常隐患数据集,并基于深度学习的目标检测技术实现了桥梁边坡常见异常隐患自动检测系统;最后将自动检测系统应用于测试集,可达到0.8492的平均精确率。实验和真实场景测试证明,与传统人工巡检方式相比,该系统具有高效性、全面性和安全性,可为公路中桥梁边坡的巡检节省大量的人力物力。

In this paper,we developed the corresponding software and hardware systems of UAV,which could realize high-precision automatic detection of those hidden perils like erosions in slope area and bridge foundation area and illegal vehicles,illegal buildings and stacked goods in the space under the bridge.First of all,we built a high-precision UAV control system,ground workstation and data management center for auto-matic acquisition of spatial images of highway bridge and slope.Then,we constructed a dataset of abnormal hidden dangers on highway,and based on the object detection technology,realized a detection system of common hidden dangers on highway.Finally,the detection system pro-posed in this paper had reached mean average precision of 0.8492 on the test dataset.Experiments and real-scene tests prove that compared with traditional manual inspections,this system is efficient,comprehensive and safe,and saves manpower and material resources for the supervision,inspection and maintenance of important construction on highway.