Seismic Damage Assessment for Wooden Detached Houses Using Drone and Deep Learning Technology in Noto

Japan is frequently struck by large-scale earthquakes, making rapid building damage assessment critically important. Conventional surveys depend on on-site inspections, which are time-consuming and constrained by road conditions. To enhance efficiency, this study employed AI and drones. A deep learning model was trained on aerial images of buildings damaged during the 2023–2024 Noto Peninsula Earthquake. Its accuracy was then validated using images excluded from the training dataset. The model’s predictions closely matched actual collapse rates, demonstrating the potential of drone-based AI for reliable and timely damage assessment.