Robust Defect Detection on PCB based on Deep Learning

S. Vijayalakshmi, K.R. Kavitha, D.P. Sangeetha, S. Thilak and G. Kavin Kumar

As a result of this advancement of the technology manufacturing printed circuit board (PCB), durable PCB defects have slowly become a censorious problem. Although PCB fault can be serious, it should not be assumed that the exact PCB distinction is a binary issue or dual-category concern. Recently, as a very popular approach called a convolutionary neural network (CNN) has become one of the deep learning frameworks, it has a majority solution in many fields, it is based on the principle of image processing, especially in image allocation or image classification. In this test case, by remodeling the fatality of this model, the multilabel learning cycle is correctly transformed into multiple labeled allocation undertaking. Then tests listed explain the desired method yields a highly reliable result due to a faults dataset.

Volume 12 | 05-Special Issue

Pages: 937-949

DOI: 10.5373/JARDCS/V12SP5/20201839