Steel Defect Detection Based on YOLOv12

This study addresses the issues of low efficiency and accuracy in traditional steel surface defect detection methods by proposing an improved YOLOv12-based detection model. The model enhances the network's feature extraction capability and recognition accuracy by incorporating the AWD module. Additionally, the Triplet Attention mechanism is integrated into the backbone network to further optimize detection performance. To improve training efficiency, the MPDIoU technique is adopted, enhancing the model's ability to handle small targets and accelerating training convergence. Experimental results show that the improved YOLOv12 model demonstrates high precision and detection speed on steel surface defect datasets, significantly improving detection efficiency and accuracy in industrial production.