Building maintenance plays an increasingly important role as buildings age. During maintenance, it is necessary to analyze building defects and record their locations when performing exterior inspections. Hence, this study proposes an automatic three-dimensional (3D) modeling method based on image analysis using unmanned aerial vehicle (UAV) flights and convolutional neural networks. A geographic information system is used to acquire geographic coordinate points (GCPs) for the geometry of the building, and a UAV is flown to collect the GCPs and images, which provide location information on the building elements and defects. Comparisons revealed that the generated 3D models were similar to the actual buildings. Next, the recorded locations of the building defects and the actual locations were examined, and the results confirmed that the defects were generated correctly. Our findings indicated that the proposed method can improve building maintenance. However, it has several limitations, which provide directions for future research.

The need for maintenance is increasing due to the aging of facilities. In this study, we proposed a crack width measurement method for images collected at safe distances using UAVs (Unmanned Aerial Vehicles). It is a method of measuring the widths of cracks using a high-resolution (VDSR) algorithm, which measures by increasing the resolution of images taken at 3 m intervals on the wall where cracks exist. In addition, the crack width measurement value was compared with a general photographed image and a high-resolution conversion image. As a result, it was confirmed that the crack width measurement of the image to which the high resolution was applied was similar to the actual measured value. These results can help improve the practical applicability of UAVs for facility safety inspections by overcoming the limits of camera resolution and distances between UAVs and facilities introduced in the facility safety inspection. However, more detailed image resolution is required to quantitatively measure the crack width; we intend to improve this through additional studies.

This study proposes an automated three-dimensional (3D)-modeling method that combines convolutional neural networks (CNNs) with unmanned aerial vehicle (UAV) technology for the efficient management of thermal anomalies in building exteriors. Conventional 3D-modeling methods for thermal imaging management either require the processing of large volumes of data due to the use of thermal distribution information from entire image regions or involve increased processing time when architectural drawings are unavailable. In this study, RGB and infrared (IR) thermal images collected via UAVs were used to automatically detect windows and thermal anomalies using a CNN-based object detection model (YOLOv5). Subsequently, Global Navigation Satellite System (GNSS)-based coordinate data and image metadata were used to convert the resolution coordinates into actual spatial coordinates, which were then vectorized to automatically generate a 3D model. The resulting 3D model demonstrated high similarity to the actual building, accurately representing the locations of thermal anomalies. This method enabled faster, more objective, and more cost-effective maintenance compared to conventional methods, making it especially beneficial for efficiently managing difficult-to-access high-rise buildings.