A Novel Deep Learning Approach to Detect Manufacturing Defects over Textile Fabric Materials

Since the evaluation of textile fabric quality is crucial to the success of every textile business, the need for computer-based intelligent visual inspection systems is on the rise. There has been a lot of interest in the study of fabric defect identification as of late due to the increasing labor costs and the increased use of automation in the textile sector. We suggested a structure based on machine learning to identify flaws in fabrics automatically. In this study, we offer an automated AI-based fabric defect identification system that makes use of deep neural network models that have already been trained. The networks are trained using pre-processed, improved versions of the fabric pictures obtained using standard image processing methods. In order to train and categorize various fabric flaws, the LightGBM is combined with a pretrained network, AlexNet. We have also examined how well our categorization model works with varying values of the experimental variables. At long last, a thorough comparison of various methods has been provided.