A Detection Approach for Wafer Detect in Industrial Manufacturing based on YOLOv8

Wafer defect detection is a critical process in semi-conductor manufacturing, ensuring the quality and reliability of chip production. In this study, a wafer defect detection network inspired by YOLOv8 was proposed. The algorithm integrates two new features, C2f and Spatial Pyramid Pooling Fusion (SPPF), to enhance the detection accuracy of the model. The C2f module improves the model's feature representation ability, while the SPPF module captures multi-scale features from input images. Experimental results on the common chip dataset WM-811K demonstrate that the proposed model outperforms Single Shot MultiBox Detector (SSD) and YOLOv5 in terms of accuracy and efficiency, achieving a mean Average Precision (mAP) of 98.9% at an intersection over union (IoU) of 0.5. This approach exhibits high accuracy and robustness in wafer defect detection and has practical value in real-world applications.