Deep Learning for Automatic Building Damage Assessment: Application in Post-Disaster Scenarios Using UAV Data
ISPRS Annals / Copernicus Publications
Ref: SRC-005-MOZ-002
Accessed: 3/30/2026
Summary
Peer-reviewed paper presenting the DEEP (Digital Engine for Emergency Photo-analysis) deep learning tool for automatic building footprint segmentation and damage classification using UAV imagery. Covers application to the 2016 Central Italy earthquake and the 2019 Cyclone Idai response in Mozambique. Co-authored by WFP's Marco Codastefano and Eoghan Bourke with Polytechnic University of Turin researchers.
View Harvard reference
Calantropio, A., Chiabrando, F., Codastefano, M. and Bourke, E. (2021) 'Deep Learning for Automatic Building Damage Assessment: Application in Post-Disaster Scenarios Using UAV Data', ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, V-1-2021, pp. 113-120. doi: 10.5194/isprs-annals-V-1-2021-113-2021.