Exploring the Role of Artificial Intelligence for Pattern Recognition of Textile Sorting and Recycling for Circular Economy
Irfan Mohammed Karmali, Omid Fatahi Valilai
Abstract
In an era where sustainability is a vital concern, the textile industry is a major obstacle that stands in the way of addressing environmental challenges, climate change, pollution, and biodiversity loss. The integration of Artificial Intelligence (AI) into the textile recycling process holds significant promise for advancing the circular economy agenda. AI technologies offer innovative solutions for automating pattern recognition tasks, thereby enhancing the efficiency and accuracy of textile sorting and recycling processes. This paper endeavors to explore AI algorithms specifically tailored for soiled cloth detection in textile products. Through a comprehensive analysis of existing literature studies, the paper aims to propose an approach that addresses the unique challenges associated with identifying and classifying soiled cloth. This research focuses on investigating and developing unsupervised computer vision techniques for the initial detection and segmentation of stains on textile materials. The paper focuses on the capabilities of two methods, (1) Otsu’s thresholding and (2) K-Means clustering, which do not require extensive training data or deep learning models. Practical conditions are examined for complex backgrounds and uneven lighting conditions and capability for accurate separation of stains from fabrics with similar colors or textures. The results show that for both methods to work efficiently, they require even and bright lighting conditions on the textiles. Moreover, the paper has discussed the challenges faced for real implementation of models for textiles being sorted on factory line.