Development of a YOLO-Based Detection Model for Key Visual Defect Elements of Tower Cranes Using UAV Images

Tower cranes are high-risk construction machines, and defects in their critical components can lead to serious safety hazards.This study developed an unmanned aerial vehicle (UAV) image-based YOLO11l object detection model for the automated visual inspection of key visual defect elements in tower cranes.A representative tower crane was selected, and 4,297 raw UAV images were obtained using multiple drone platforms.After image filtering and synthetic image generation, a dataset of 350 images was constructed, including 229 real UAV images and 121 synthetic images.Seven classes related to mast bolts and bolt caps, turntable motor boxes, and main jib stopper packing were annotated with bounding boxes.On the test dataset, the model achieved precision, recall, mAP@50, and mAP@50-95 values of 0.849, 0.779, 0.878, and 0.670, respectively.These results validate the feasibility of UAV-based automated defect detection for tower crane inspection.