Real Time Defect Identification of White Fabric in Textile Industry using Computer Vision

Abstract

This study focuses on the defect identification of white fabric in the textile industry based on quality control standards. The standard manual process for examining fabric defects are labor and cost expensive. In this study, a high-quality camera will be used to capture the image of the fabric moving from the conveyor belt that will be processed to identify defects such as horizontal or vertical stripes, yarn missing, bunching up, and stains. Comparative analysis of the machine learning technique will be performed to find the best method. The method will be evaluated on a dataset captured through the local textile industry. At the production level, if some defect is found the machine will stop working unless that defect is resolved.