An Improved Lightweight YOLOv11 Algorithm for Weld Surface Defect Detection

Industrial welding often exhibits some essential problems, such as unclear defect characteristics and complex backgroundinformation. However, the existing defect detection models have relatively high costs and may be weak in weld defect detection.To address the problem, this paper proposes an improved lightweight YOLOv11 model for welding surface defect detection,called YOLO-Air. First, the model integrates the feature extraction module with the convolutional module to enhance featureextraction capabilities and optimize computational resource utilization. Second, the GSConv and VOV-GSCSP modules areembedded in the neck network to reduce feature redundancy in both spatial and channel dimensions, and then lower thecomputational load. Third, a lightweight detection head is designed as part of the detection network to further reduce modelcomplexity. Finally, we compare our proposed YOLO-Air model to the baseline model via the Welding Defect Test-V2 andNEU-DET dataset. Experimental results show that the proposed model achieves superior weld surface detection effects,especially the metric mAP50 is +1.3%, the parameter number is reduced by 17.3%, and the computational load is reduced by 31.7%