One of Italy’s largest apple suppliers has added AI-based checks to its ripeness testing process

Italian apple association Vip says it has welcomed a valued new team member to its quality department: artificial intelligence.

Its role? To identify the perfect time for harvesting.

For just over a year, in fact, the company has been experimenting with an AI-based system that can review slices of newly picked fruit and – based on an extensive database of previous tests – and tell if they are sufficiently ripe.

It’s the latest step in a process of digitisation that Vip’s quality department has undergone in a relatively short space of time.

Around five weeks before each of its different apple varieties are picked, the group’s member growers submit thousands of individual fruits for testing.

Each one is assessed to determine when those still on the trees will reach the ideal level of ripeness for harvesting.

As Wolfgang Graiss, head of Vip’s quality control department, explains, this “is crucial “for both flavour and the optimal storage that enables us to supply customers consistently all year round”.

His colleaue André Trafoier says one of the main methods of assessment is based on starch content.

To measure this, apples are sliced and sprayed with an iodine-potassium solution, to which they react: the lighter the slices remain, the higher the starch content and the less ripe the apple.

Vip AI apple testing 2

Manual, digital, artificial

Until a few years ago, such observations were carried out with the naked eye, and values were calculated by hand.

Over time, however, the process became a digital one: nowadays, software can calculate average values, while a camera captures and saves photos of each slice, thus ensuring everything is traceable and well documented.

Then in October 2023, VIP started training AI to compare the starch levels gauged by employees with those it suggested.

In the months leading up to its first use, the system was fed photographs of over 27,000 apples of differing varieties and maturity levels.

Trafoier notes: “The results were excellent for varieties and ripeness levels with extensive data available.”

After carefully considering proposals from software developers in New Zealand and Canada, Vip says it found an ideal partner called Fuxware in Naturno, just 12km away from its own headquarters, and where its major cooperatives are based.

“Choosing to work with Fuxware turned out to be the right decision from the start,” says Andreas Oberhofer, Vip’s head of IT. “They worked incredibly quickly and are always on hand to answer our questions.”

The experience has been so positive, in fact, Vip says it has gained valuable insights into AI that it is keen to share with other companies considering similar projects.

AI, it seems, offers a helping hand that may well extend beyond Vip’s own operations to the rest of the apple industry, and possibly even further.

Vip AI apple testing 1