Surface Defect Detection Method of Lead Frame Based on Knowledge Distillation

As the carrier of integrated circuit, semiconductor lead frame is an important part of chip electrical connection. In the manufacturing process, the lead frame must go through multiple processes, which will lead to defects on the surface. However, it is difficult to obtain defect data in actual manufacturing. In this paper, an algorithm for surface defect detection of lead frame based on knowledge distillation is proposed. We propose an unsupervised model learning strategy. The algorithm uses normal images for training and uses global context information and local texture information for defect detection. The experimental results show that the proposed method can effectively detect the surface defects of the lead frame. This method can not only detect structural defects such as dirt, scratches, and pin deformation, but also detect logical defects such as plating area deviation.