Sometimes all it takes is fresh eyes to see clearly what has been in front of you all along.
In mid-December, Yakima, Wash.-based grower Apple King updated the Ellips apple grader used to categorize its fruit to a new Ellips “deep learning” sorting model called TrueAI. According to Kyle Mills, Apple King’s operations manager, his team has already seen outstanding results.
“We have always felt that our Ellips grading technology provided us with a competitive advantage based on its overall performance, including the ability to accurately detect both external and internal defects,” Mills said in a news release. “With the TrueAI upgrade, we have significantly enhanced our grading capabilities. We are now able to detect even the most difficult defects, such as stem bowl cracks, at a very high level of accuracy with minimal false positives. During the first three days of operation, we ran Honeycrisps and galas, both of which exhibited a fair amount of stem bowl cracks, and our results were truly impressive.”
Related news: Shorter Washington apple crop brings challenges to marketers
Three years in the making, Ellips’ TrueAI model delivers high-level grading accuracy by detecting complex defects and produce characteristics, according to the release. In 2022, TrueAI was implemented across multiple commodities such as dates, pomegranates, apples and most recently, onions.
New categories that the TrueAI system will incorporate include blueberries, cherries, kiwis and other fresh fruits and vegetables, according to the release. What’s more, as grading performance moves towards 100% detection accuracy and becomes more automated, grading and sorting tasks can be done with less staff.
“From day one, the improvements were immediate and visually apparent. Even apples with difficult-to-detect defects that previously might have ended up in the packing area were correctly sorted,” Ray Keller, owner of Apple King, said in the release. “This improved performance will help in our continual quest to improve productivity while delivering the most consistent pack to our customers. Additionally, we can now more effectively incorporate automated packing technology at the various packing exits since manual intervention can be virtually eliminated.”
The deep learning AI technology can be added to new or existing systems and requires little system operator training or effort, and TrueAI updates, based on targeted defects or specific needs, can be incorporated into the existing system, according to the release.
