New Wind Turbine Dataset Advances Automated Coating Defect Detection

Researchers developed CoatingDet, a publicly available dataset containing 5,416 high-resolution images of wind turbine tower coatings captured under real manufacturing conditions, with annotations for critical defects and benign surface particles. Validation with YOLOv11n and RT-DETR showed the dataset can support reproducible deep learning-based inspection while highlighting the importance of lighting conditions and class imbalance for reliable deployment.