DisasterBench: A Multimodal Benchmark for UAV-Based Disaster Response in Complex Environments
Researchers have introduced DisasterBench, a new multimodal benchmark designed to evaluate AI models in complex disaster response scenarios using UAV imagery. This benchmark covers 14 disaster types and 9 critical tasks, focusing on reasoning capabilities like causal attribution and decision-making, rather than just perception. Alongside the benchmark, they developed DisasterVL, a lightweight 2B-parameter multimodal model that demonstrates competitive performance against larger models, achieving GPT-4o-level accuracy with greater efficiency. AI
IMPACT Enhances AI capabilities for critical disaster response tasks, potentially improving efficiency and accuracy in real-world emergencies.