Defective Product Classification System for Smart Factory Based on Deep Learning

Smart factories merge various technologies in a manufacturing environment order to improve factory performance and product quality. In recent years, these smart have received lot of attention from researchers. this paper, we introduce defective classification system based on deep learning for application factories. The key component the proposed is programmable logic controller (PLC) artificial intelligence (AI) embedded board; call an AI Edge-PLC module. A pre-trained model uploaded cloud service where can access download it use certain product, case, electrical wiring. Next, setup collect wiring data real-world environment. Then, applied preprocessing collected extract region interest (ROI) images. Due limitations availability appropriate labeled data, used transfer method re-train our purposes. models were then optimized applications boards. After carrying out tasks, wire dataset previously published casting dataset, using neural networks including VGGNet, ResNet, DenseNet, GoogLeNet, analyzed results achieved by system. experimental show that able classify products quickly with high accuracy