Researchers have developed HASTE, a no-code web platform designed for rapid post-disaster building damage assessment. HASTE enables non-machine learning experts to create damage maps from satellite imagery within hours of a disaster. The platform employs two methods: one that trains a semantic segmentation model on user-labeled data from a single scene, and another that uses pretrained vision models and logistic regression for quick assessments. Preliminary experiments show HASTE's effectiveness, matching supervised baselines with significantly less data, and it has already supported over thirty real-world disaster responses. AI
IMPACT This platform could significantly speed up disaster response by providing critical damage assessments within hours.
RANK_REASON The cluster describes a research paper detailing a new platform for disaster assessment.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
