A Brief Survey on Fabric Defect Detection

Textile Industry with constant growth over the years have evolved from hand woven to machine fabricated fabrics. This large scale production of fabrics by machines, which in turn became source to variety of defects from various phases like production, dying and printing, became costly and difficult to be managed manually for quality assurance. Therefore, a thorough review is essential to comprehend how Fabric Defect Detection patterns have changed over time. The research presented in this review offers a feature-focused systematic survey. This review categorizes the fabric defect detection methods into Manual and Auto-Derived features. Manual feature extraction methods discuss traditional defect detection algorithms proposed based on the expert knowledge of the researcher and industry operator. Whereas in Auto derived feature extraction-based methods various state of the art deep learning algorithms are discussed in the field of Fabric Defect detection. An insight on public fabric defect detection datasets and ideas (remark) on how to enhance models by bridging the gaps is also shared.