ACN-YOLO: Solar cell surface defect detection based on improved YOLOv8s

This paper presents a solar cell surface defect detection method based on the improved YOLOv8s algorithm, termed ACN-YOLO. The method significantly enhances the detection of micro-defects on solar cell surfaces by incorporating the AKConv module, the C2f-Faster-EMA attention mechanism module, and the nwd_loss function module. The AKConv module utilizes irregular convolution techniques to enhance the global feature representation of samples, while the C2f-Faster-EMA module improves the detection capability for small target defects, and the nwd_loss function optimizes detection accuracy for tiny objects. Experimental results demonstrate that ACN-YOLO achieves an mAP@0.5 index of 84.7% with an increased detection speed of 295.3 FPS, meeting real-time monitoring requirements. The improved YOLOv8s model shows increases of 16.7%, 14.6%, and 11.8% in precision, recall, and mean average precision, respectively, compared to the original model, validating the effectiveness and practicality of the proposed improvements.