Automating Postearthquake Recovery in Construction: Leveraging BIM, Deep Learning, and Web Map Services for Efficient Solutions

Abstract

In response to the increasing complexity encountered in postearthquake recovery within the construction sector, there is a pressing need for more accurate and efficient methods. Leveraging emerging technologies to automate processes is a feasible approach to tackle these challenges effectively. This study focuses on integrating building information modeling (BIM), deep learning (DL), and web map services (WMS) to accelerate and improve the recovery planning for earthquake-damaged buildings. This integrated approach not only supports sustainable construction goals but also has the potential to reduce planning time by 10%–15%, depending on project situations, while decreasing the resources required. The research used a BIM-based platform, applying a vision transformer for image classification and Detectron2 for instance segmentation, to accurately identify damage in structural elements and streamline damage evaluation process through automation. Implementing a BIM- and WMS-based plugin further enhances this process, enabling automated and data-driven recovery planning incorporating sustainability considerations. A case study demonstrated this integrated BIM-DL-WMS approach on an actual recovery project. The result underlines the potential of these technologies to revolutionize postearthquake recovery efforts in the construction industry. They make the recovery process more efficient, accurate, and sustainable.