PCB Board Defect Detection Algorithm Based on Improved YOLOv8

To address the issues of insufficient fine-grained defect feature extraction and limited detection accuracy in traditional YOLOv8 algorithm models for PCB industrial board defect detection, this paper proposes a PCB board defect detection method based on an improved YOLOv8 network. This method introduces partial convolutional modules and multi-scale channel attention mechanisms into the head network of the original YOLOv8 algorithm framework to enhance the model’s feature extraction capabilities and multi-scale information fusion capabilities for defect regions. The PConv module effectively improves the model’s perception of local features in defect regions, while the MLCA mechanism enhances the model’s ability to focus on features across different scale channels, thereby improving detection accuracy and robustness. Experiments show that the improved algorithm model achieves an average accuracy of 87.3% and a recall rate of 82.1%, which are 4.1% and 2.9% higher than the original YOLOv8 model, respectively. This study provides an efficient and high-precision solution for PCB board defect detection, which is of great value for practical industrial detection.