Abstract
After the typhoon hit, if we could quickly carry out damage assessment for residents, repair of the house and apply of disaster insurance could also be provided sooner than now. Based on the deep learning method, the drone photos of the Kyonan Town, which were taken one year after Typhoon Faxai, were used to detect and evaluate the roof and wall damage of houses, in this paper. This paper is mainly divided into four parts: taking photo using drone, outputting the texture of the roof and wall, detecting damage from the texture and classify the level of damage of the houses. The average F value of the detection was more than 0.80 and classification of that was about 0.65.
