DisasterRoute improves drone image answers with fast accurate checks
DisasterRoute: Evidence-Guided Dual-System Routing for Visual Disaster Question Answering
Computer Vision and Pattern Recognition
Summary
Answering questions about disaster scenes from drone images needs to be quick and correct to help responders. The authors created Disaster-S1, a tool that quickly picks answers by looking at the picture and question once. They also made DisasterRoute, which decides when to trust the quick answer or use a slower, smarter review to get a better response when needed. This balance helps save time while making sure answers are more accurate during disasters.
What this means in practice
- •For disaster management teams: Use fast and accurate question answering from drone images to support quick decisions on disaster sites with limited computing power.
- •For mobile robotics developers: Integrate adaptive decision systems that combine fast direct answers with selective detailed review to optimize robot perception tasks.
Authors
Tan Zhang, Xing Deng, Lu Zhang, Xiaofeng Zhu, Ping Hu
Abstract
Visual question answering from low-altitude UAV imagery must deliver timely and accurate judgments to support disaster response under limited edge computing resources. Routine questions may be answered through direct decisions, whereas more challenging questions may benefit from step-by-step reasoning. However, existing research primarily focuses on improving models' visual understanding and generative reasoning capabilities, with limited exploration of efficient direct decision-making and its adaptive coordination with generative reasoning. We develop Disaster-S1 based on a vision-language model adapted to the disaster domain. Trained for option discrimination, it outputs a probability distribution over answer options in a single multimodal forward pass for fast decision-making. We further propose DisasterRoute, a dual-system routing framework that combines prediction confidence with disaster evidence to decide whether to accept the initial answer or invoke the disaster reasoning model DisasterVL for review. By allocating inference computation on demand, the framework retains the efficiency of direct decision-making and improves overall question-answering accuracy through selective review. Experiments on DisasterBench show that Disaster-S1 achieves superior accuracy with low latency, while DisasterRoute further improves accuracy through selective review at substantially lower latency than full reasoning.