Imaging Recognition of Earthquake Disaster with Digital Twins
指導老師：陳俊杉 教授
專案經理：張慰慈、林彥廷
地震不同於颱風、海嘯、水患、山火等其他類型的災害，它非但難以預判與預警，且會在極短時間內帶來大範圍、全面且毀滅性的傷害。基於地震勘災與救災的需求，我們需要在極短的時間內判定災損的目標（如房屋、橋梁、道路）位置與可能的經濟損失、受困人數等。航照與勘災影像的解析度較高，搭配3D地理資訊系統（3D GIS）的模型成像與機器學習技術便可快速找出受災目標，彙整回3D GIS系統後就能進一步計算出災損的經濟規模、受影響人數與其他潛在的地震災損資訊，供救災單位參考。
Different from the other kinds of disasters, such as typhoon, tsunami, flood, and wildfire, earthquake is almost impossible to be predicted and early warned. Furthermore, it causes extensive, composite and devastating damage in a very short period of time. Search and Rescue (SAR) team has to determine attacked targets, such as buildings, bridges, and roads, and the degree of damage as soon as possible for the following activities. Base on such requirement, we plan to apply machine learning technology on analyzing the videos collected by Unmanned Aerial Vehicle (UAV) with digital twins from 3D Geographic Information System (GIS) system for quickly recognizing these damaged objects in earthquake and extracting some hided information for the SAR team.