Overview

HASTE (High-speed Assessment and Satellite Tracking for Emergencies) is an AI-powered tool from the Microsoft AI for Good Lab for quickly assessing building damage after a disaster. You add satellite imagery, label a small amount of it, and HASTE predicts damage across the whole area.

Choose your path

HASTE offers two workflows that both turn imagery into a damage assessment — pick based on what you need and how fast you need it.

	Rapid Building Assessment	Damage Mapping
How it works	Embed buildings → label a few → an in-browser model predicts the rest → validate → report	Draw labels → train a segmentation model → predict across the whole image
Output	Per-building damaged/intact, with accuracy metrics and a damaged-building estimate	A continuous damage map + downloadable geopackage
Speed	Minutes — no training job	Slower — GPU model training and inference
Needs	Building footprints for the area	Just imagery

Not sure? Start here.

If you have building footprints and want an answer fast, use Rapid Building Assessment — it’s the quickest route from imagery to a damage assessment. Choose Damage Mapping when you need a continuous, pixel-level damage raster.

Both workflows start the same way — create a project and add an image layer — then diverge. Those shared building blocks have their own pages: Projects, Image layers, and the Model catalog.