“Exploring the Role of Artificial Intelligence for Pattern Recognition of Textile Sorting and Recycling for Circular Economy” is a peer-reviewed research article by Irfan Mohammed Karmali and Omid Fatahi Valilai. The article examines how artificial intelligence (AI) can be applied to automate and improve pattern recognition tasks in textile sorting and recycling, with the aim of advancing circular economy practices within the textile industry.  

Key Takeaways

• Textile sustainability challenge: The study emphasises that the textile industry represents a significant obstacle to environmental sustainability due to its resource intensity and waste generation, making improved recycling processes essential.  
• AI for textile recycling: AI technologies, particularly computer vision algorithms, are explored as tools to automate pattern recognition in textile products, which can improve the efficiency and accuracy of sorting for recycling purposes.  
• Focus on soiled cloth detection: The paper specifically investigates AI approaches tailored to detect and classify soiled cloth in textile streams a challenging task due to varying textures, colours and lighting conditions.  
• Methods analysed: Two unsupervised computer vision methods are evaluated: Otsu’s thresholding and K-Means clustering, chosen because they do not require large, labelled training datasets or deep learning models.  
• Lighting and real-world conditions: The research finds that both methods perform best under even and bright lighting, highlighting practical challenges for deployment on production lines where lighting and background conditions vary.  
• Implementation barriers: While the AI methods show promise in controlled conditions, the article discusses challenges to real-world implementation, particularly in industrial environments with complex visual conditions.  

This study demonstrates that AI-driven pattern recognition has potential to enhance textile sorting and recycling, a key component of circular economy strategies for the fashion and textile industry. By analysing unsupervised techniques that can operate without extensive training data, the research highlights both opportunities and practical limitations for deploying these approaches in industrial settings. Continued exploration of AI methods adapted to real-world conditions could support more efficient textile recycling systems and contribute to broader sustainability goals.

Read the entire article on ScienceDirect.