"Optical sorting is the future," begins Bram Jansen, Sales and Export Manager at Schouten, which is part of the Invaro Group. This Dutch company focuses on grading and processing machine development and design, for among other products, (seed) potatoes. He sees, especially, new software availability and technology like deep learning - where, for example, potatoes are used to train the machine - accelerates the optical quality sorting process.
Bram notes that labor plays a significant role when choosing optical graders. "Workers are becoming increasingly scarce. It's almost impossible to get people, and size or quality sorting by hand is boring work. What's visible to the human eye, a camera can see, too. Hopefully, in the future, sitting in sorting rooms will become obsolete," he says.
Bram says optical size and quality graders are replacing human labor and are, in part, ready to do final checks, too. "You can, thus, save several people in the sorting room. Eventually, you'll need only one person for post-checking and to oversee the process." Though he expects, in time, that, too, will be unnecessary for most standard varieties or batches with no unusual flaws.
Capacity is another essential factor. The 12-track optical size grader Optica CS, for instance, sorts out clods and growth defects, does square-size grading, and can grade up to 18 tons of potatoes per hour into nine different sizes. Bram points out that optical size grader is far more accurate than mechanical size sorting.
"And that immediately represents a revenue model. With mechanical grading, oversized seed potatoes are often sorted out and considered less valuable consumption potatoes. Optical size grading gives you the use of more seed potatoes from your gross batch. Tests show optical graders sort up to 25% more accurately," says Jansen, calculating that optical size sorting can save up to €500 per hectare per year compared to mechanical sorting.
Schouten has been marketing optical size graders for ten years, and the company recently developed an optical quality grader. The software was being used for flowers, but it was adapted and made suitable for (seed) potatoes. They tested a prototype, and the Optica Q is now ready to be put into service as a zero series.
Schouten has also begun - with a customer panel - developing an optical quality grader for potatoes that you can add to an existing sorting line. Bram says capacity proved to be a significant consideration for clients. "On average, a person can assess five to seven tons of potatoes per hour; using optical quality grading, we boost that to ten to 15 tons per hour. That can easily save two employees, and you can use post-checking staff more efficiently," he says, explaining the machine's revenue model.
The company used deep learning to train the optical quality grader. "We took pictures of a sample of about 20 potatoes affected by, say, scab, for example, where the flaw is marked with a color. That was added to the database. Eventually, the optical grader itself recognizes the potatoes with the scab." It can distinguish defects, including Rhizoctonia, wireworm, grubbing damage, or green discoloration. And, Bram points out, if, in a given season, the potatoes have unique or new quality flaws, the graders can easily be 'tutored'. "The deep learning software lends itself well to making the machine suitable for other products, like, say, onions, too," he concludes.