Machine learning offers solution to cut textile dyeing waste

A recent study suggests that machine learning can help lower textile production waste by better predicting how fabric colours shift during the dyeing process.

Warren Jasper, a professor at the Wilson College of Textiles in Raleigh, North Carolina, explains that fabrics are dyed when wet, but the final shade is only visible once they’re dry. Since colour shifts during drying, any mistake in dyeing isn’t noticed until later, by which time more fabric has already been processed. This delay leads to unnecessary waste in production.

The colour change from wet to dry fabric isn’t consistent across shades. This non-linear pattern means each colour reacts differently, making it hard to apply data from one sample to another.

To solve this issue, Jasper created five machine learning models, one of which was a neural network built to handle complex, non-linear data. He trained these models using visual information from 763 dyed fabric samples in both wet and dry states. As each dyeing session took hours, gathering this data was a time-intensive process.

All the machine learning models performed better than traditional methods, but the neural network proved to be the most accurate. Using the CIEDE2000 colour difference formula, it recorded an error as low as 0.01 and a median error of 0.7, well within industry standards, where values above 0.8 to 1.0 are often seen as unacceptable. Other models ranged from 1.1 to 1.6, while the non-ML baseline reached a high of 13.8.

By offering accurate colour predictions, this neural network could help fabric producers avoid costly errors and reduce waste, making it possible to spot issues before dyeing large quantities.

“We’re slightly behind in adopting new technology in textiles,” Jasper says. “Although the industry is slowly beginning to use machine learning, the progress is slow. These models could play a big role in reducing waste and boosting efficiency, especially in continuous dyeing, which makes up more than 60% of all dyed fabrics.”