HASTE Platform Speeds Disaster Damage Mapping

Researchers at Microsoft developed HASTE (High-speed Assessment and Satellite Tracking for Emergencies), a web platform that allows analysts to create per-building damage maps from satellite imagery following a disaster. The platform requires minimal machine learning expertise and operates using two methods.

One method trains a model on user-labeled polygons within a disaster scene. The other utilizes pre-trained vision models and logistic regression, requiring labeling of only a few buildings.

HASTE has already supported over thirty real-world disaster responses since 2023, including earthquakes, hurricanes, and wildfires. The system delivers damage assessments to humanitarian organizations within hours or days of imagery being captured.

The developers plan to expand HASTE’s capabilities to include damage assessment for infrastructure like roads, and are exploring vision-language models and active learning techniques to improve accuracy and efficiency. The platform is available as an open-source project.