SHINTARO GOTOH
Department Department of Environment Systems, Faculty of Geo-Environmental Science Position Professor |
|
Language | Japanese |
Publication Date | 2021/05/20 |
Type | |
Title | Research on early detection of total damage in the disaster using AI |
Contribution Type | |
Journal Type | Japan |
Details | The disaster area of Typhoon19th was detected by UAV and from the ortho image of the disaster situation and the point cloud data, the AI model was used to examine the model of total destruction and partial destruction and the damage situation detecting model.
For this object, we created 3D images from the extraction of damage to houses and public facilities, and data for verification of drawings for formulating reconstruction plans using only point clouds. In 2D, YOLO and SSD were applied as a model for the damage detection of public facilities and the disaster area was extracted and the accuracy was evaluated from the image of the disaster area. |