LUCAi™ for Stonefruit

How LUCAi™ achieves 99%+ detection accuracy for stonefruit defects

Traditional machine learning approaches to fruit grading work by analyzing small sections of the fruit, a few pixels at a time and making local predictions. It can't fully assess how those pixels relate to each other across the whole fruit.

LUCAi™ sees the whole fruit. It uses a deep learning classification model, designed to mimic the human brain's visual cortex, which is the part of the brain that processes what we "see" with our eyes. This allows LUCAi™ not just to recognize individual defects from the images produced by Spectrim but also make a decision based on the overall appearance and context of the entire piece of stonefruit.

This advanced model allows LUCAi™ to correctly identify and categorize previously hard-to-detect stonefruit defects such as rot, split pit, scars, spurs, softness and physical damage, and many other common defects.

When analyzing each fruit, LUCAi™ will assign:

1.  Defect Class/Category: Identifies the specific type of defect (e.g., rot, split pit, spurs, etc.).
2. Confidence Score: Indicates how certain LUCAi™ is that a defect has been correctly classified.
3. Severity Rating: Provides a 0-100 measurement of how severe the defect is, based on its size, color contrast, and spread. This gives operators clear control over how tightly or loosely they wish to sort the fruit based on quality thresholds.