Benchmark RadarGeoDisaster
GeoDisaster is an operational geospatial disaster reasoning benchmark with 2,921 instances across 43 question types and five task families, integrating EO/GIS evidence and grounding answers in executable geospatial workflows.
- Released
- 2026-06-15
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
GeoDisaster addresses the gap in evaluating tool-grounded spatial reasoning and structured decision-making for disaster response, offering a potential standard for assessing operational geo-intelligence in RS-VLMs and agentic systems.
Motivation
Remote-sensing vision-language models (RS-VLMs) have advanced Earth-observation analysis toward visual interpretation and instruction-following, yet fall short of operational geo-intelligence, which demands tool-grounded spatial reasoning and structured, evidence-backed decisions.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.