Colour recognition in smart packaging

A study conducted at Finland’s University of Vaasa could lead to smart packaging that indicates product condition through colour-changing printing inks.

Doctoral researcher Jari Isohanni has investigated how machine learning can most effectively be utilised in creating smart packaging that recognises colours. Combined with machine-based recognition of subtle colour changes, printed packages with colour-changing inks could offer a cost-effective solution for industries such as food and beverage, health care, logistics and electronics.

According to Isohanni, up to this point there has been a lack of research on which colour recognition method should be used and when. His research fills this gap by comparing the applicability of machine learning and colour difference methodologies to different situations.

“My research showed that traditional, simple computational methods work well for recognising significant colour differences. However, for subtle changes and varying conditions, the most effective methods were convolutional neural networks that are based on artificial intelligence,” Isohanni explained.

The right method for the right purpose

Functional printing inks change colour according to conditions — for example, when temperature rises or humidity increases. The colour change in printing ink is so subtle or rapid that it cannot be recognised effectively enough with current machine vision methods, Isohanni said. By the time the ink’s colour change is mechanically detectable, the process may already have progressed too far or damage may have occurred.

This is why research focused on detecting small colour changes opens up new possibilities for industry compared to current electronic sensors.

AI enables automatic colour recognition in industry with nearly human-eye accuracy, providing new tools for quality control, among other applications. Based on the research, solutions can also be developed for consumers to indicate the condition of food products or other goods. The ink indicator used in the doctoral research can be printed on packaging alongside regular labels, with minimal additional cost compared to electronic smart packaging solutions, according to Isohanni.

Expensive electronic measuring devices cannot be placed on, for example, a lettuce package, as this would constitute a large portion of the product’s price or could cause additional challenges for recycling. Printed indicators solve this problem, Isohanni said.

Though it’s still at the dissertation stage (with examination happening in early September), Isohanni’s research could pave the way for an environmentally friendly alternative that can improve processes and enhance consumer information — for example, by monitoring shelf life in the food industry.

Image credit: iStock.com/guteksk7