Design of a Real-Time Warp and Weft Measurement System for Fabric Analysis in Textile Testing Using Machine Learning

The terms warp and weft are most commonly used in woven fabric construction. Woven fabrics are made by interlacement of warp and weft which are made up of two set of yarns with the longitudinal / lengthwise / vertical set of yarns called warp and horizontal / transverse set of yarns called as weft. Number of threads per inch of warp and weft determine the compactness of a fabric and weight of fabric. A lot of research is in progress towards automating the warp and weft threads in linear distance measurement with most of them using pre-recorded images. In this paper, design and evaluation of a real-time warp and weft measurement system for textile testing using computer vision and machine learning algorithms is proposed. The machine learning algorithm employed for proposed work is an unsupervised K-means clustering algorithm. The performance of proposed system is evaluated on two types of silk sarees, with and without identical warp and weft yarn colour in the saree. The proposed system costs only $32 with an overall response time varying between 1.42–3.09s based on the type of saree and hardware used. The proposed work can be easily employed in various textile industries to reduce the burden of humans involved in manual warp and weft threads per inch measurement for quality checks in textile testing.