Can We Predict the Unpredictable? Leveraging DisasterNet-LLM for Multimodal Disaster Classification

TL;DR

DisasterNet-LLM is a novel multimodal large language model that significantly improves disaster classification accuracy by integrating images, weather data, and reports, aiding timely disaster response.

Contribution

The paper introduces DisasterNet-LLM, a specialized LLM with cross-modal attention and adaptive transformers for comprehensive disaster analysis, outperforming existing models.

Findings

Achieved 89.5% accuracy in disaster classification

Attained an 88.0% F1 score, demonstrating balanced precision and recall

Reached 0.92 AUC, indicating strong discrimination ability

Abstract

Effective disaster management requires timely and accurate insights, yet traditional methods struggle to integrate multimodal data such as images, weather records, and textual reports. To address this, we propose DisasterNet-LLM, a specialized Large Language Model (LLM) designed for comprehensive disaster analysis. By leveraging advanced pretraining, cross-modal attention mechanisms, and adaptive transformers, DisasterNet-LLM excels in disaster classification. Experimental results demonstrate its superiority over state-of-the-art models, achieving higher accuracy of 89.5%, an F1 score of 88.0%, AUC of 0.92%, and BERTScore of 0.88% in multimodal disaster classification tasks.