Site verification tests for UAV bridge inspection and damage image detection based on deep learning

摘要：

　　With the fast spreading use Unmanned Aerial Vehicle(UAV)in aerial photography and successful application of Convolutional Neural Network based image processing algorithms,Deep Learning,in many fields such as image classification and object detections,the possibility of combine these two technologies to a revolutionary bridge inspection approach becomes more and more clear.In this study,the feasibility of bridge inspection using UAV is investigated by a series of bridge site verifications.And bridge damage photo reported in recent bridge inspection conducted in Japan are used to train some Convolutional Neural Network models,and the accuracy of the damage detection machines are evaluated.Some technics to train a high accuracy model with some data based are also tested and the results are also reported.

关键词： UAV Convolutional Neural Network Bridge Inspection Deep Learning Damage Detection

作者: J.Dang A.Shrestha D.Haruta Y.Tabata P.Chun K.Okubo

作者单位: Department of Civil and Environmental Engineering,Saitama University,Saitama,Japan Department of Civil and Environmental Engineering,Ehime University,Matsuyama,Japan

会议类型: 国际会议

会议名称: The 7th World Conference on Structural Control and Monitoring(7WCSCM)(第七届结构控制与监测世界大会)

会议地点: 青岛

会议语种:英文

页码: 1548-1557

在线出版日期: 2018-07-22（万方平台首次上网日期，不代表论文的发表时间）