When a disaster occurs, it is important to know as soon as possible how many people are affected, where they live, and the severity of the damage, to properly plan response operations.

The Automated Damage Assessment tool (ADA) is a deep learning model that identifies damaged buildings in satellite images after a disaster has happened. Automating this assessment makes it faster and far less dependent on human effort, therefore enabling 510 to quickly deliver the information that is needed. The assessment time decreases from weeks to hours, which makes a difference in the speed of operations and, ultimately, on the number of lives that can be saved.

“I found ADA to be very helpful in terms of getting an initial understanding of the situation. By using AI-assisted satellite damage tools, we were able to notice areas that were highlighted in multiple models and on-the-ground data to help us determine which version was likely to be the most accurate.” 

Corey Dickinson, Data Consultant at IFRC