Computer Vision Based Research on PCB Recognition Using SSD Neural Network

Abstract

Printed circuit board (PCB) is an important component in the information technology industry, during PCB board assembly process, jack localization and recognition is particularly important. In view of the problem that the accuracy of target detection is not ideal, this paper applies deep learning convolutional neural network framework SSD (single shot multibox detection) to PCB element detection. By obtaining a large number of original PCB images, creating semantic label image dataset, feeding them into neural networks, and adjusting the hyperparameters, the best classification accuracy and localization accuracy are achieved. At the same time, the advantages of different algorithms are analysed after tests with other deep learning network methods. Experiments show that the SSD deep learning algorithm can realize accurate detection process of jack localization and classification, which is better than other algorithms.