A Multimodal, Multilingual, and Multidimensional Pipeline for Fine-grained Crowdsourcing Earthquake Damage Evaluation

TL;DR

This paper introduces a structured pipeline using multimodal large language models to analyze social media data for rapid, fine-grained earthquake damage assessment, demonstrating promising results across multiple events.

Contribution

It presents a novel multimodal, multilingual, and multidimensional pipeline leveraging MLLMs for disaster impact assessment from social media data.

Findings

MLLMs effectively integrate image-text signals for earthquake damage evaluation.

Strong correlation observed between MLLM assessments and ground-truth seismic data.

Performance varies with language, distance, and modality.

Abstract

Rapid, fine-grained disaster damage assessment is essential for effective emergency response, yet remains challenging due to limited ground sensors and delays in official reporting. Social media provides a rich, real-time source of human-centric observations, but its multimodal and unstructured nature presents challenges for traditional analytical methods. In this study, we propose a structured Multimodal, Multilingual, and Multidimensional (3M) pipeline that leverages multimodal large language models (MLLMs) to assess disaster impacts. We evaluate three foundation models across two major earthquake events using both macro- and micro-level analyses. Results show that MLLMs effectively integrate image-text signals and demonstrate a strong correlation with ground-truth seismic data. However, performance varies with language, epicentral distance, and input modality. This work highlights the potential of MLLMs for disaster assessment and provides a foundation for future research in applying MLLMs to real-time crisis contexts. The code and data are released at: https://github.com/missa7481/EMNLP25_earthquake