A Deep Learning Artificial Neural Network Based Approach to Automated Fabric Fault Detection

Detection of fabric faults performs a significant part in textile production. Detection of fabric defects is a very challenging task because of the wide variety of defects that occur in fabrics. Due to the defects in the fabric, 45-65% of income are lost by manufacturers. Fabric Defect Detection (FDD) is done by manual inspection method and it is very tedious and consumes more time. An Automated fabric fault identification system, is a specific method mainly designed to focus on finding fabric defects. Such a system is designed to assure quality of the fabrics produced. To identify and classifythe defects in the fabrics, an architecture based on Deep learning Artificial Neural Network is proposed in this paper. Convolutional Neural Network (CNN), with multiple convolution and pooling layers, is proposedto increase the efficiency and accuracy of classification of defects in fabrics.