Improvement of the defect inspection process of deteriorated buildings with scan to BIM and image-based automatic defect classification

Currently, safety inspections of aging buildings are conducted by inspectors manually recording defects identified through visual inspection of paper blueprints, which carries the risk of data loss and potential issues with the reliability of the inspection results. This paper proposes an integrated defect information management process that utilizes a CNN model and Scan to BIM technology to solve these problems. This study developed a CNN-based deep learning model that detects defects in the exterior walls of aging buildings using image data and classifies them into five types. Additionally, BIM scanning was implemented using a laser scanner to acquire shape information from existing buildings without blueprints and digital data. A Construction Operations Building Information Exchange (COBie) file was defined for managing defect data, with a Dynamo script developed to automatically link the stored information to Scan to BIM. A case study was conducted to examine the feasibility and efficiency of a BIM defect information management model developed in the field. The proposed process was applied to four aging buildings, with the results confirming its efficiency in terms of time and cost. The results of this study are deemed appropriate for the integrated management of safety inspection data and are expected to reduce the workload of inspectors and support efficient maintenance decision-making. •Improvement measures for issues in the safety inspection process of deteriorated buildings.•Safety inspection of building exteriors using artificial intelligence and 3D modeling technology.•Utilizing Scan to BIM and COBie files for defect information management.•Integrated management of safety inspection data and decision-making support tools for maintenance.