This study developed a building exterior tile detachment detection system that integrates drones, infrared thermography, and the YOLO model. The system utilizes a drone equipped with a Raspberry Pi and a dual-light gimbal camera (infrared and visible) to inspect building facades, accurately identifying hollowed and exposed cement areas. Infrared thermography is applied to detect abnormal temperature regions on walls, and the YOLO model is used to identify and mark exposed cement areas. The annotated images are analyzed and overlaid for risk assessment, and evaluation reports are then generated. Additionally, the system incorporates 3D modeling technology to reconstruct building exteriors. By integrating the 3D model with risk assessment reports, the system effectively highlights areas of tile detachment risk, enhancing users’ grasp and monitoring of building maintenance. This system not only enhances detection accuracy and efficiency but also significantly reduces time and labor costs, offering reliable data support for building upkeep.