Building damage assessment in natural disasters: A trans- and interdisciplinary approach combining domain knowledge, 3D machine learning, and crowdsourcing

Recent natural disasters have claimed many lives. Reliable damage predictions and timely assessments are essential for effective rescue operation planning and efficient allocation of limited resources. Currently, experts in the field perform damage assessment manually, which is resource- and time-intensive. To address this issue, we propose a general trans- and interdisciplinary concept that combines the strengths of domain knowledge, automated computational methods, and crowdsourcing. The objective is to provide relevant and timely damage information after a natural disaster. The specific implementation presented for the earthquake damage use case includes (1) the development of a set of novel, innovative methods, (2) their combination to obtain timely and reliable damage information, (3) fully defined interfaces between all components to ensure an automated data flow, (4) implementation as a fully open-source framework, and (5) the participation of end users in the development of the framework from the beginning, contributing their expertise. Compared to other existing individual solutions, our interdisciplinary implementation has shown to provide fast and accurate information in disaster situations, aiding the management of consequences and saving lives. We consider the implementation transferable to various types of natural hazards due to its open-source realisation and the flexibility of its modules and interfaces.