Damage Detection and Level Classification of Roof Damage after Typhoon Faxai Based on Aerial Photos and Deep Learning

Following the occurrence of a typhoon, quick damage assessment can facilitate dispatch house repair and disaster insurance works. Employing deep learning method, this study used aerial photos Chiba prefecture obtained following Typhoon Faxai in 2019, to automatically detect evaluate roof damage. This comprised three parts: training model, detecting using trained classifying level The detection object outline, blue tarps, completely destroyed roof. roofs were divided into categories: without damage, with destroyed. F value proposed method was higher than those other methods. In addition, it be further five levels from 0 4. Finally, spatial distribution analyzed ArcGIS tools. is expected provide certain reference for real-time after typhoon.