PCB surface defect detection method based on improved YOLOv8 algorithm

The invention discloses a PCB surface defect detection method based on an improved YOLOv8 algorithm. According to the method, an improved model based on YOLOv8 is adopted for defect detection; in the model, an up-sampling process is added in a Neck network of YOLOv8, an original FPN and PAN structure is replaced by a BiFPN structure, and a multi-level feature network is obtained by using weighted feature fusion and cross-scale connection; a C2f-CAM module is provided, the CAM attention mechanism is fused into the C2f module by the C2f-CAM module, and the C2f module in the up-sampling process in the Neck network is replaced by the C2f-CAM module; meanwhile, when the model is trained, a normalized Wasserstein distance optimization loss function is introduced; the method can improve the detection precision of PCB surface defects, has high accuracy and recall rate, and solves the problem that the YOLOv8 algorithm is poor in effect in the small target detection process. 本发明公开了一种基于改进YOLOv8算法的PCB表面缺陷检测方法；该方法采用基于YOLO