Comparative Analysis of Existing Deep Learning Techniques for Automatic Defects Detection in Printed Circuit Boards

Printed Circuit Board (PCB) defect identification is an crucial task in the electronics industry, as it may help in production of high quality product. Deep learning-based techniques have shown promising results in detecting PCB defects. Several automated defect detection techniques have recently been used by numerous industries for the early detection of any PCB manufacturing errors. Such businesses commonly utilize deep learning-based image recognition techniques for the same. In this research work., we have discussed recent PCB defect detection algorithm based on deep learning concepts. We have also gone through numerous publicly accessible industrial PCB image datasets. A modified YOLOv5s model has also been proposed specifically used for PCB image datasets, with the available defect detection techniques. Important parameters including Accuracy, Sensitivity, Specificity, and AUC have been used to compare performance.