14 May 2025More than a buzzword, Artificial Intelligence (AI) is transforming the direction of optical sorting. Bühler’s groundbreaking new machine, SORTEX AI700, has ‘ripped up the script’ and completely altered how an optical sorter detects impurities, resulting in the best defect concentration ever, and more quality and profits for food processors.
Every industry, from construction to healthcare, is talking about AI. But how does it apply to agriculture, and in particular, food processing, and why should processors take notice?
Melvyn Penna knows a thing or two about optical sorting. Recently celebrating his 35th year with SORTEX, he has played a pivotal role in the advancements of the optical sorter. Recently, he mentioned that he is working on the most exciting project yet. The following is an interview with Melvyn, focusing on the topic of AI and its direct impact upon food safety and quality.
What is AI in optical sorting?
Melvyn: AI is a broad term which encompasses many levels of computer learning. In the context of optical sorters, in the past we have had machine learning, where the machine is set up by an engineer and the learning algorithm tries to improve on the initial set up. This is an effective form of AI seen in our SORTEX SpectraVision series.
We are now implementing a different type of AI, or what we call ‘deep learning’. Here, the machine uses CNN, or Convolutional Neural Networks, to train the algorithm to make its sorting decisions. The machine is trained with millions of labelled images, each one giving the algorithm greater understanding about what is accept and what is reject. Like a human brain and eye, it uses all the available information, colour, shape, and texture, to evaluate the product.
How does this directly benefit food processors?
Melvyn: With this technology implemented in our new SORTEX model, the AI700, we have found improved defect removal efficiencies whilst rejecting less good product. This translates to processors getting more good product from the same amount of incoming product, enabling processors to take more to market, reduce waste, and earn more profit. In a real-world example, over a period of a year, 5-10% better reject concentration can translate to hundreds of thousands in revenue that would have been lost.Page 2
Did any future trends trigger this development?
Melvyn: The future presents global challenges and trends in food processing, which if left unaddressed, will damage both processors and consumers. These trends include the rising costs of raw materials, climate change and the increasing prevalence of defects, increasingly stringent market regulations, and higher consumer standards and awareness.
Another important trend right now is allergen detection. SORTEX AI700’s first application will be Oats. Here, we have achieved new levels of defect removal, correctly identifying barley, wheat and rye from oats. This is vital, due to oats being gluten free and the others being allergenic. Traditional sorters, while still achieving regulation standards, do not achieve the same level of gluten removal in oats as the SORTEX AI700.
Alongside allergens, the AI700 addresses the cost of raw materials by maximising yield, helping solve the prevalence of defects, overcoming market regulations, and ensuring consumers are safe.