Hyperspectral imaging & AI sorting: enabling scalable textile recycling
The textile industry has long been a major contributor to global waste, with an estimated 92 million tons of textile waste generated each year. While recycling efforts have made progress, one of the greatest hurdles remains sorting textiles accurately and efficiently. Traditional methods are often inefficient, relying on manual labor or basic mechanical processes, both of which are time-consuming and prone to errors. In response to this, a new wave of technology is emerging, with hyperspectral imaging and AI sorting systems at the forefront of improving textile recycling. But do these technologies really hold the potential to scale recycling efforts and create a more sustainable industry, or is it just another overhyped solution?
The promise of Hyperspectral imaging and AI
Hyperspectral imaging is a technique that captures a wide spectrum of light beyond the visible range, allowing for the identification of materials based on their unique spectral signatures. When applied to textile recycling, hyperspectral imaging can precisely detect the composition and type of fabric, whether cotton, polyester, wool, or blends. This is crucial for recycling because different fibers require different processing methods. AI sorting, which uses machine learning algorithms, can then take this data to automate and optimize the sorting process, ensuring that the right textiles are sent to the appropriate recycling streams.
The combination of hyperspectral imaging and AI holds substantial promise for textile recycling. For one, it addresses one of the main issues in the industry: the accurate sorting of textiles. Traditional sorting methods, which often rely on visual inspection or basic mechanical separation, are inefficient and lead to contamination, ultimately lowering the quality of recycled materials. A 2025 systematic review “Fashion and textile waste management in the circular economy compiled and analyzed 104 research papers on textile waste generation and recycling globally (or across many countries). The authors’ note that of the textile waste generated each year (estimated globally at ~ 92 million tonnes) about 50 % is post‑consumer waste. They report that only 15–20 % of waste is collected for reuse or recycling; the rest (i.e. “more than 80 % of clothing waste”) is discarded to landfill or incineration, a direct result of inefficient sorting practices. Hyperspectral imaging, by providing precise data about fabric composition, significantly improves sorting efficiency and can drastically reduce contamination in the recycling process.
Moreover, AI sorting enhances this by learning from past sorting data to continuously improve its accuracy. Unlike human workers, AI systems can analyze vast quantities of data in real-time, making them much faster and more efficient than traditional methods. PICVISA, a textile‑sorting/ recycling machinery company, reports having built a “fully automated” textile‑sorting machine using the Specim FX17 hyperspectral camera (900–1700 nm NIR range), coupled with AI/optical separation and side‑blowing hardware, capable of sorting textiles by composition (cotton, polyester, viscose, etc.), color, and shape. The system is reportedly used at a plant in A Coruña, Galicia, processing ~5,000 tonnes of textile waste per year. A new study “Hyperspectral imaging quantifies blend composition change in workwear textiles” examines the use of hyperspectral data to quantify fiber blend composition, helping address a core challenge: many garments are neither pure cotton nor pure polyester but blends. Their findings support the feasibility of spectral methods to detect and distinguish blends
The Challenges: Cost, Scalability, and Technological Limitations
Despite the clear advantages, there are substantial challenges in the widespread adoption of hyperspectral imaging and AI sorting in textile recycling. One of the primary concerns is the cost of implementing such technologies. While the technology has shown promise in pilot projects and small-scale applications, scaling it to handle the vast quantities of textile waste generated globally is a significant financial investment. The infrastructure required for hyperspectral imaging systems, coupled with the need for high-powered AI systems, and could deter many recycling facilities, especially those in developing regions, from adopting these solutions.
In addition, the technology itself is still in its early stages, and its effectiveness can be impacted by certain factors. Hyperspectral imaging can struggle with distinguishing between similar materials, especially in the case of textiles that are heavily dyed or blended. The process of sorting mixed fibers remains a challenging task, as some fibers may appear visually similar in the spectral signature but require very different recycling methods. This limitation can result in inefficiencies, reducing the overall effectiveness of the system.
Furthermore, AI systems require large, high-quality datasets to function properly. While AI sorting systems improve over time, the initial setup and training process can be labor-intensive. In order to function at their best, AI systems must be trained on diverse and representative datasets of textile materials. Without access to these large, high-quality datasets, the AI may not be able to optimize the sorting process as intended, leading to potential errors or missed opportunities for material recovery.
Addressing the challenges
While these challenges are legitimate, they are not insurmountable. On the cost front, the technology is likely to become more affordable as it matures and adoption increases. The initial high investment in infrastructure can be offset by the long-term economic benefits of higher-quality recycled materials and reduced waste sent to landfills.
On the technological side, advancements in hyperspectral imaging and AI are progressing rapidly. AI’s ability to adapt and learn from new data means that the system can continue to improve its accuracy over time. Additionally, collaborations between textile manufacturers, recyclers, and tech companies are working toward creating more robust AI models that can handle mixed fabrics and difficult-to-recycle textiles. The creation of large, comprehensive textile datasets is an area of active research, with some companies already making strides in compiling diverse textile libraries to train AI systems more effectively.
Additionally, advancements in hyperspectral imaging sensors are helping to overcome some of the limitations with mixed fabrics. New technologies, such as multispectral sensors, are being developed to better handle the complexity of textile compositions, improving the reliability of sorting systems even when dealing with complex or heavily processed textiles.
The integration of hyperspectral imaging and AI sorting into textile recycling has the potential to revolutionize the way we manage textile waste. While there are obstacles to overcome, the evidence suggests that these technologies are not just a passing trend but a viable solution for the future of textile recycling. As research continues and the technology becomes more accessible, the adoption of AI and hyperspectral imaging will likely become more widespread, helping to scale up recycling efforts and move toward a more sustainable fashion industry.
The key to realizing this potential will be collaboration. Governments, industries, and tech companies must work together to overcome the financial and technical challenges. With continued investment, research, and scaling, hyperspectral imaging and AI sorting could become a cornerstone of a sustainable circular economy for textiles. The road may not be easy, but it is clear that these technologies are paving the way for a more sustainable and efficient recycling system.
