Grain grading is built on both visual and non-visual assessments. One of the most consistently important visual grading factors across crop types is broken kernels. Whether you’re evaluating wheat, lentils, or soybeans, broken kernels impact everything from processing efficiency to overall grade.

At Ground Truth Ag, our benchtop MV/NIRS system uses high-resolution machine vision to detect and quantify broken kernels in real time, bringing greater consistency and speed to a process that’s often subjective and time-consuming.

What Are Broken Kernels?

Broken kernels are exactly what they sound like: pieces of grain that have fractured, chipped, or split during harvest, handling, or storage. They may be caused by mechanical damage, crop condition at harvest, or post-harvest handling.

While some degree of breakage is common, excessive broken kernels reduce the usability of the grain and often result in downgrades during official grading.

We currently detect and quantify broken kernels in:

Canadian Western Hard Red Spring Wheat – Broken kernels can impact flour yield, dough consistency, and overall grade.

Red Lentils – Broken lentils affect dehulling rates, marketability, and export value.

Soybeans – Breakage can lower crush efficiency and increase storage losses.

Support for Canadian Western amber durum, oats, faba beans are in development as we expand our grain models.

Why Do Broken Kernels Matter?

For Farmers & Sellers – High rates of broken kernels can trigger downgrades or price discounts. Identifying this early helps make informed marketing and storage decisions.

For Buyers & Processors – Broken kernels impact end-use value, especially in milling, dehulling, and crushing. Consistency in detection helps align supply with buyer requirements.

For Quality Control – Breakage levels are a defining characteristic in visual grading and directly influence contract compliance and bin management.

How We Detect Broken Kernels

Detection of broken kernels is powered by machine learning models trained on thousands of annotated samples across different crops, grades, and conditions. By learning what breakage looks like from real-world examples, the system can reliably identify broken kernels based on shape, size, and structural features.

This approach helps ensure consistency across samples, operators, and locations, removing much of the subjectivity that traditionally comes with visual inspection; which often looks at only a portion of the sample. Our system evaluates the entire sample, helping capture a completer and more representative picture of grain quality.

With real-time detection of broken kernels, users can:

Spot high breakage levels earlier

Avoid surprises during delivery or inspection

Make more confident decisions around sorting, binning, or blending

When paired with all other detection models, this feeds into a full grading and quality assessment that goes beyond just counting defects and supports decisions you can act on.

Meet the Factor: Broken Kernels - A Key Visual Indicator of Grain Quality

Meet Ni Nguyen