DAMAGE LEVEL PREDICTION OF BRIDGE MEMBERS USING DEEP LEARNING WITH DRONE CAMERA PICTURES.

Abstract

Japanese bridges are aging, there are not enough civil engineers who inspect the bridge at the city hall. So there is a need for an inspection method that can easily grasp the deterioration of bridges. Therefore, we invented a method that can be easily inspected even by staff who are not familiar with the civil engineering field. First, this method uploads an enlarged photo of the bridge taken with a digital camera mounted on a drone etc. The result of inferring the degree of damage from the uploaded photo by deep learning is automatically output. So that even municipal office staff who are not expert with civil engineering can easily handle it. I invented a method that automatically outputs by deep learning. As a result, photos were uploaded locally from smartphones and tablet terminals, and the results of bridge deterioration were output instantly, making it possible to diagnose bridge deterioration without expertise in civil engineering.