Multilevel Deep Learning Model for Fabric Classification and Defect Detection

Abstract

Detecting defects in fabrics is a difficult task as there are a lot of variations in the type of fabric and the defect itself. Many methods have been proposed to solve this problem, but their detection rate and accuracy were very low depending on the model tested. To eliminate the variations and to improve the performance, we implemented multilevel modeling in our approach. This article proposes an enhanced and more accurate approach to detecting tissue defects. Here, we compare the performance of various advanced deep learning models such as MobileNetV2, Xception, VGG19, and InceptionV3 and how their performance changes with the type of fabric. First, a Convolutional Neural Network model is used to classify the fabric into different types with an accuracy of 97.6%, and then on the basis of the type of fabric, the best model is used to detect defects in the fabric. This has a significant advantage in improving the overall performance of fabric defect detection. In addition, K-fold cross-validation has been carried out to verify the coherence of the proposed model.