YOLOv11-FST: An Improved YOLOv11-Based Model for Solar Photovoltaic Cell Defect Detection

Currently, defect detection in photovoltaic cells suffers from challenges such as difficulty in identifying small defects and ambiguous defect features. To address these issues, this paper proposes a deep learning-based photovoltaic cell defect detection method. FasterBlock modules are first embedded into the YOLOv11 backbone to boost feature extraction with fewer parameters, followed by replacing the original neck with SlimNeck to lower complexity without sacrificing accuracy. Additionally, a TripletAttention module is added to the small-object detection layer to better identify tiny defects. Experiments show the improved network achieves mAP50 scores of $\mathbf{9 2. 3 \%}$ and $\mathbf{6 2. 5 \%}$, exceeding YOLOv11s by $\mathbf{1. 4 \%}$ and $\mathbf{1. 0 \%}$, highlighting its effectiveness for photovoltaic defect detection