Multimodal Support System for Disaster Risk Management

2026-04-12 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Advanced, medium

Summary

NOAH is a computational multimodal system designed to support disaster risk management (DRM) in Brazilian cities by facilitating information exchange between public DRM agents and at-risk populations. Developed using artificial intelligence, NOAH integrates a chatbot with natural language processing (NLP), speech recognition, image classification, and retrieval-augmented generation (RAG). The system communicates directly with the population via WhatsApp, collecting reports in Portuguese across text, audio, and image formats. NOAH employs BERTopic for text classification, Whisper Small for audio transcription, and Resnet50 convolutional neural networks for visual incident analysis. This approach aims to provide a practical and scalable tool for municipal Civil Protection and Defense agencies, enhancing decision-making and emergency response efficiency in Portuguese-speaking regions.

Key takeaway

For research scientists developing public safety systems, NOAH demonstrates a robust multimodal AI architecture for disaster risk management. You should consider integrating diverse AI components like NLP, speech recognition, and image classification to process varied input from the public. This approach, particularly using platforms like WhatsApp, can significantly improve data collection and decision support for emergency services, especially in specific linguistic contexts like Portuguese-speaking regions.

Key insights

NOAH is a multimodal AI system for disaster management, integrating NLP, speech, and vision via WhatsApp for Brazilian cities.

Principles

• Multimodal input enhances disaster reporting.
• AI can scale disaster response tools.
• Direct communication improves information flow.

Method

NOAH combines BERTopic for text classification, Whisper Small for audio transcription, and Resnet50 for image classification to process multimodal disaster reports from WhatsApp.

In practice

• Use WhatsApp for direct public communication.
• Integrate BERTopic for topic modeling.
• Apply Resnet50 for visual incident analysis.

Topics

• Disaster Risk Management
• Multimodal AI System
• Natural Language Processing
• Retrieval-Augmented Generation
• BERTopic

Best for: Research Scientist, AI Scientist, NLP Engineer, Computer Vision Engineer

Related on AIssential

• Disaster Question Answering with LoRA Efficiency and Accurate End Position — A LoRA-optimized Japanese BERT-Bi-LSTM hybrid system delivers high accuracy and efficiency for disaster-specific question answering.
• Build an Emergency Helpline Voice Agent with LangChain — A low-latency, voice-only AI emergency agent can triage and dispatch services using concurrent STT, LLM, and TTS processing.
• DisasterBench: A Multimodal Benchmark for UAV-Based Disaster Response in Complex Environments — Multi-stage multimodal reasoning is crucial for effective UAV-based disaster response, requiring specialized benchmarks and lightweight models.
• MAviS: A Multimodal Conversational Assistant For Avian Species — Domain-adaptive multimodal LLMs are crucial for fine-grained understanding in specialized ecological applications like avian species.
• Helping disaster response teams turn AI into action across Asia — AI offers critical tools for enhancing disaster response efficiency and effectiveness in resource-constrained environments.
• Paper Digest: ICASSP 2026 Papers & Highlights — ICASSP 2026 highlights showcase advancements in signal processing, AI, and multimodal learning, emphasizing efficiency and robustness.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.