SHINTARO GOTOH
   Department   Department of Environment Systems, Faculty of Geo-Environmental Science
   Position   Professor
Language English
Publication Date 2022/09/30
Type
Peer Review Peer reviewed
Title Disaster Area Detection by Deep Learning using UAV and Satellite Imagery
Contribution Type
Journal Proc. of Asian Conference on Remote sensing 2022
Journal TypeAnother Country
Publisher Asian Conference on Remote sensing
Responsible for 論文統括、結果の評価
Authorship Corresponding author
Author and coauthor Kazuaki AOKI, Shintaro GOTO. Chitomi SAKAI
Details Due to recent changes in the global environment, damage from disasters such as extreme
weather, typhoons, and heavy rainfall has been occurring frequently. In this study, we investigated a
method for automatic detection of damaged areas from aerial photographs taken by UAV and
satellite photographs in order to obtain detailed information on the damage at an early stage. We
applied a deep learning-based method to detect damaged areas from orthomosaic images of
disaster-stricken areas using data taken by a drone in the area affected by Typhoon 19 and other
disasters. The construction of a system that automatically detects damaged areas is expected to lead
to an early understanding situations and cost reductions.