Fault Detection and Quality Inspection of Printed Circuit Board Using Yolo-v7 Algorithm of Deep Learning

The requirements of the contemporary manufacturing environment where the delivery of 100% defect-free PCBs is expected, have increased the significance of the printed circuit boards (PCBs) inspection process. Billions of Electronic products are manufactured annually, and the success rate of proper working is 97 % out of 100%. The remaining 3% is faulty products and most faults occur due to PCBs. This is a huge loss for the company; it is therefore needed to overcome the problem. This research will conduct a study of the newest model YOLO v7 (You-Only-Look-Once) algorithm of deep learning to find out the solution to minimize the loss of the company, it is an advanced kind of image classification in which an end-to-end neural network identifies defects in an image and highlights them with bounding boxes. This work is presented for the quality inspection, different types of fault detection, and classification of PCBs. Deep learning algorithms, such as convolutional neural networks (CNN), due to their high accuracy and efficiency have achieved considerable attention. In this proposed approach a highly accurate dataset was taken from The Open Lab of Peking University. The data set includes 1386 images having six kinds of defects (open circuit, spur, short circuit, missing hole, mouse bite, and spurious copper). This research aims to bring a solution not to have faulty PCBs and further decrease the manufacturing cost and product waste and enhance the manufacturing process of the company.