Microsoft’s AI helps pinpoint earthquake damage in Myanmar for faster disaster relief

Microsoft’s AI-powered satellite imaging technology is helping disaster response efforts in Myanmar by rapidly assessing earthquake damage. By combining aerial imagery with specialized computer vision models, the company’s AI for Good Lab identified over 500 severely damaged buildings in Mandalay following a devastating 7.7 magnitude earthquake. This technology provides critical information to aid organizations, helping them prioritize resources and target relief efforts more effectively in the crucial early hours after natural disasters.

The big picture: Microsoft’s philanthropic AI for Good Lab analyzed satellite imagery to assess building damage in Mandalay following Friday’s 7.7 magnitude earthquake that struck Myanmar.

• The analysis identified 515 buildings with 80-100% damage and another 1,524 structures with 20-80% damage, providing crucial targeting information for relief organizations.
• While the assessment requires on-ground verification, the data has already been shared with aid groups including the Red Cross to guide initial response efforts.

Behind the numbers: The technology combined satellite imagery from Planet Labs with customized AI computer vision models specifically built for Mandalay’s architecture and landscape.

• Standard AI models can’t effectively analyze diverse disaster scenarios because “the Earth is too different, the natural disasters are too different and the imagery we get from satellites is just too different to work in every situation,” according to Microsoft’s chief data scientist Juan Lavista Ferres.
• While tracking destruction from certain disasters like wildfires follows predictable patterns, earthquakes present unique challenges because they affect entire urban areas in complex ways.

The challenges: Morning cloud cover initially prevented the technology from capturing clear imagery immediately after the earthquake struck.

• The team had to wait for clouds to move before a San Francisco-based Planet Labs satellite could capture usable aerial pictures.
• Once images were obtained, they were transmitted to Microsoft’s headquarters in Redmond, Washington, where a team was waiting to process the data late Friday night.

Why this matters: Rapid damage assessment is crucial in the critical first hours following natural disasters when responders need to prioritize limited resources.

• The technology pinpoints specific locations of damage, allowing response teams to more efficiently direct personnel and supplies to the hardest-hit areas.
• This approach represents an evolution of the AI for Good lab’s previous work tracking disasters like Libya’s 2023 flooding and Los Angeles wildfires.