A New Approach for Classification of Different Woven Fabric Patterns and Thread Densities With Convolutional Neural Networks

Abstract

Fabrics produced from microfilaments are superior to conventional fiber fabrics, due to their properties such as light weight, durability, waterproofness, windproofness, breathability and drapeability. Tightly woven fabrics produced from microfilament yarns have a very compact structure due to small pore dimensions between the fibers inside the yarns and between yarns themselves. It is almost very difficult to distinguish the structures of densely woven fabrics with the visual evaluation. Therefore, it is very important to automatically determine the differences in the texture properties of such fabrics. Thanks to the developments in image acquision technology and image processing methods, the texture classification of fabrics can be estimated more quickly and reliably than visual inspection. In this study, the classification of high-density microfilament woven fabrics according to different texture types and thread density was achieved by using the ResNet-50 algorithm. The obtained results were evaluated in a confusion matrix form. The classification accuracy of the CNN algorithm was measured as 0.95 on average.

Optimization of Process Parameters Based on Machine Learning and Determination of Relative Importance of Process Variables on Jute Ply Yarn Breaking Extension

“…The algorithm works by recursively partitioning training data into smaller subsets based on threshold values of decisive features. Decision trees with several terminal nodes are created using the CART algorithm, allowing for parameter optimization [11,12].…”