ChangeQuery: A Multimodal AI Framework for Post-Disaster Remote Sensing Analysis

Researchers have introduced ChangeQuery, a unified multimodal framework designed to improve post-disaster situational awareness by integrating pre-event optical satellite imagery with post-event synthetic aperture radar (SAR) structural features. The framework addresses limitations in existing vision-language models, which rely on unimodal optical data and exhibit a bias toward natural disasters, neglecting human-induced events like armed conflicts. To support this, the team created the Disaster-Induced Change Query (DICQ) dataset, a large-scale benchmark that balances natural catastrophes and armed conflicts. ChangeQuery aims to move beyond pixel-level change detection to high-level semantic understanding, enabling more actionable intelligence for complex strategic queries in disaster response. The work is detailed in a preprint on arXiv (ID: 2604.22333).