The rapidly increasing number of aged buildings has been in the blind spot of building maintenance-related laws for a long time. Recently, the need for building maintenance due to the collapse of old buildings has emerged, but there is a very lack of professional inspection personnel for safety inspection and diagnosis. This study proposes a smart safety inspection and maintenance system using advanced IT equipment and deep learning technology to efficiently manage existing old buildings, which can be divided into five stages.

In the first stage, images and 3D point cloud data are collected and matched using drones and laser scanners for digital transformation of existing buildings. In the second stage, a CNN-based deep learning model is proposed to detect defects using image data from buildings. The CNN model analyzes the image data of the acquired old building to automatically detect damage and deterioration areas of the building and classify the types of defects. In the third stage, a BIM model using 3D point cloud data is generated. The BIM model through reverse engineering can visually check the shape information and defect location of the actual building. In the four stage, a maintenance information management plan based on the BIM model was proposed. In this study, we proposed a COBie file for defect data management, which can predict the repair priority and cost of defects according to the importance of defects. In the five stage, the effectiveness of the proposed smart maintenance process was verified through the case study. Through this, it was confirmed that the proposed method is easy to identify defect information that occurs during actual safety inspection and diagnosis work and can support continuous maintenance in one file.

In the future, the proposed smart maintenance process can serve as a kind of construction database and is expected to be used as a tool for communication between stakeholders and efficient decision support for building maintenance.