Objective An improved YOLOv8 algorithm for photovoltaic panel defect detection is proposed to address the problems of low detection accuracy high computational cost and a large number of parameters in existing photovoltaic panel defect detection algorithms. Methods First the improved C2f-MS module was used to replace a part of the original C2f structure in the model. This not only reduced the computational cost and the number of parameters of the model but also enhanced the extraction and fusion capabilities of multi-scale features. Second the normalized Gaussian Wasserstein distance NWD was added to the original CIoU to improve the detection performance for small targets making the model?? s detection ability for various targets more balanced. Finally Soft-NMS was used to replace NMS which changed the way the original model processed prediction boxes solved the problem of multiple detection boxes for a single target and further improved the detection results. Results Experimental results showed that the improved YOLOv8 model reduced the number of parameters by 9. 57% and the computational cost by 6. 1%. Meanwhile the mAP@ 50 increased from 87% to 89. 5% an increase of 2. 5% and the mAP@ 0. 5 ∶ 0. 95 increased from 45. 7% to 49. 8% an increase of 4. 1% . Conclusion The improved model shows a certain improvement in detection accuracy while reducing the number of parameters and computational cost which has a certain reference value for practical applications.