Stones & soil clods
Mineral foreign material close in size to the seed, removed on color and contrast.
Food and agriculture
MAS AS Series Beans & Seeds Sorter
Agricultural seed streams vary by season, origin and crop year, so natural variation can look a lot like a defect. This page defines what to inspect — and why the accept / reject classes have to be confirmed on your real samples.
Color gradingShape & sizeForeign material
Material types
Each seed stream is its own recipe.
Color, size and surface differ across crops, so the accepted-product window is set per stream. Click a type to see what defines it.
Kidney beans
Color & skinDeep red beans where color uniformity, skin damage and discoloration are the main grading targets. Splits and wrinkled skins are common reject classes.
Visible defects
What the sample test should classify.
Each class is defined and confirmed on real seeds — not assumed from generic wording.
Stems & plant matter
Pods, stems and chaff carried in from the field that must leave the accepted stream.
Mould & discoloration
Visibly mouldy, stained or weather-damaged seeds graded out for food quality.
Broken & split
Splits, halves and chipped pieces separated when whole-seed product is the target.
Off-color seeds
Greens, darks or bleached pieces outside the accepted color window for the crop.
Foreign seeds & debris
Other crop seeds, glass, plastic or packaging debris defined as reject during testing.
Accept / reject
From mixed harvest to a graded stream.
Sorting is defined by what stays in the accepted product and what is pulled into the reject — confirmed on representative samples.
FeedMixed seed feed
Incoming product carries good seeds plus stones, stems, off-color and broken pieces.
Illustrative only. Real grading depends on seed color, size, surface condition, presentation and the confirmed recipe.
Why it needs testing
Natural variation is the hard part.
Seed streams change with season and origin, so a recipe is built from evidence, not assumptions.
Variation looks like a defect
Natural color and shape spread across a crop year can resemble a defect — the accepted window has to be tuned on real seeds.
Small foreign material
Stones and seeds close to product size often need a dedicated sample test to confirm reliable separation.
Recipe tied to real examples
AI-assisted recognition stays anchored to verified accepted and rejected examples rather than open-ended claims.
How it works
From spread to graded reject.
A controlled, sample-led loop tuned to the accepted and reject classes you confirm in testing.
- 01Customer-prepared
Material spread review
Seeds are metered and spread into an even layer so every piece can be inspected without overlap.
- 02Class definition
Define visible classes
Accepted color, size and surface ranges plus the named reject classes are set from your real samples.
- 03Controlled inspection
Optical & multi-view inspection
Lighting and cameras read color, shape and surface — with multi-view used where seeds present unevenly.
- 04Air ejection
Reject inspection & ejection
Off-class pieces are ejected and the reject stream is reviewed so good product is not lost.
- 05Handoff
Operator handoff
Recipe, accepted window and reject review are handed to the operator after the result is confirmed.
Inspection path
Technologies behind the grade.
Beans and seeds are typically a visible-color and shape problem, supported by AI-assisted recognition and multi-view inspection.
Real sample testing
Confirm your bean and seed grade with real samples.
A beans-and-seeds project should start from your real stream, including the accepted product and the off-color, broken and foreign pieces that must be removed.
Request a Real Material TestOnline preview
Preview your beans and seeds stream online first.
Upload sample photos for a visual accept / reject preview before sending real material. Real sorting performance still needs a machine test.
Start AI Visual Preview