Scalable Approach to Create Annotated Disaster Image Database Supporting AI Driven Damage Assessment

Abstract

This work proposes an AI-driven system for enhancing hurricane damage estimation through the use of deep learning models for precise identification and Classification of damaged building components. The system incorporates high-resolution aerial and satellite images to create an annotated database to enhance data analysis and processing. A CNN-based approach detects and classifies structural damage with the ability to distinguish between minor and major impact categories. The system employs geospatial data for precise localization and real-time alarm systems to aid emergency response teams. Through damage evaluation automation, this work aims to accelerate disaster response, reduce human effort, and improve recovery planning.