Abstract: The horizontal hidden-frame glass curtain wall is a common form of building envelope. The absence of lower-edge brackets directly affects the structural safety of the curtain wall during the operation and maintenance phase. This paper focuses on the brackets of the horizontal hidden-frame glass curtain wall and proposes a non-destructive detection method for missing brackets. Unmanned aerial vehicles （UAVs） equipped with visible-light sensors and infrared thermal imaging cameras are employed to collect data on these brackets， enabling the 3D modeling of building facades for analysis. The deep learning target detection algorithm YOLOv8 is used to detect the bracket targets in infrared thermograms of the glass curtain wall. A dataset is formed for the screened missing results of the glass curtain wall brackets， allowing defects to be visually positioned in the three-dimensional model and retrieved at any time. An application report for detecting missing brackets can be generated to document their locations and quantities， forming a complete inspection workflow—from data collection and processing to visualization—through a visualization software platform. This system helps assess the health status of the curtain wall structure， supports its operation and maintenance， and extends its service life.