Drone-Based Imaging and Automated Classification for Glass Facade Condition Monitoring

Glass facade inspections are traditionally timeconsuming, costly, and unsafe, especially for high-rise buildings. This research proposes an automated system that integrates drone-based imaging, image stitching, and YOLO-based object detection for real-time monitoring of glass facades. The DJI Neo drone captures high-resolution images of the building facade, which are then stitched together to create panoramic views. YOLO models (v8, v11, v12) are used to detect defects such as cracks, chips, and discolorations. YOLOv12 outperformed YOLOv8 and YOLOv11 in defect detection, achieving $70 \% \mathrm{mAP}$ on the test set, demonstrating superior precision and recall for real-time facade monitoring. This study provides a cost-effective, scalable, and safer solution for facade inspection compared to traditional methods.