Circuit Manufacturing Defect Detection Using VGG16 Convolutional Neural Networks

Manufacturing, one of the most valuable industries in world, is boundlessly automatable yet still quite stuck traditionally manual and slow processes. Industry 4.0 racing to define a new era digital manufacturing through Internet Things- (IoT-) connected machines factory systems, fully comprehensive data gathering, seamless implementation data-driven decision-making action taking. Both academia industry understand tremendous value modernizing are pioneering bleeding-edge strides every day optimize largest world. IoT production, functional testing, fault detection equipment already being used today’s maturing smart paradigm superintend intelligent perform automated defect order enhance production quality efficiency. This paper presents powerful precise computer vision model for classification product from standard product. Human operators inspectors without aid must spend inordinate amounts time poring over visual data, especially high volume environments. Our works quickly accurately sparing defective entering doomed operations that would otherwise incur waste form wasted worker-hours, tardy disposition, field failure. We use convolutional neural network (CNN) with Visual Geometry Group 16 layers (VGG16) architecture train it on Printed Circuit Board (PCB) dataset 3175 RBG images. The resultant trained model, assisted by finely tuned optimizers learning rates, classifies 97.01% validating accuracy.