Open EO for accessible, automated disaster damage assessment
2021-09-30, 09:30–10:00, Group on Earth Observations

In recent years, the development and deployment of machine learning models for structural damage assessment after natural disasters has been recognized as an innovative way to tackle the increased frequency and intensity of these devastating events brought on by climate change. For example, convolutional neural networks are trained on multitemporal satellite imagery (pre- and post-disaster) to harness change detection and output the damage levels incurred on each building and develop saliency maps to demonstrate which areas were affected the most severely. Now more than ever, it is important that the latest technological tools, particularly those that are artificial intelligence-based, are made available to local communities, individuals, nonprofit organizations, and local governments in disaster-prone, isolated, and underserved areas. Open earth observation enables the dissemination of the end products of the high-level research conducted in this area. Real-time EO data, in combination with the corresponding ML models, can be deployed in mobile apps, for example.

Some related work I've done in this field is here:
However, the focus on this talk would be more on deployment and open EO rather than technicalities.

Please see the abstract above.

Authors and Affiliations

Thomas Y. Chen, Academy for Mathematics, Science, and Engineering


Transition to FOSS4G


FOSS4G implementations in strategic application domains: land management, crisis/disaster response, smart cities, population mapping, climate change, ocean and marine monitoring, etc.


1 - Principiants. No required specific knowledge is needed.

Language of the Presentation