Fruit sorting trials are often discussed as a question of whether a machine can recognize a defect, grade, color condition, or internal-quality target. Before that question can be answered fairly, the trial team needs an agreed way to assess the fruit itself. Reference labels provide that foundation.
A reference label is the documented assessment attached to a fruit or sample group before the sorting result is reviewed. It may record a measured value, an agreed appearance category, a visible condition, an internal reference result, or another stated assessment. It is not a performance claim; it shows what the trial result is compared with.
Without it, a disagreement can become confused: the issue may be an equipment decision, or the fruit may have been classified incorrectly in the first place. A usable record keeps those questions separate.
Start With the Decision the Label Must Support
The label should be designed around a real grade or handling decision, not a broad description such as “good fruit” or “defective fruit.” State the condition, commercial consequence, and expected route.
For example, a packhouse may need to separate:
- fruit that meets an agreed premium-pack appearance standard;
- fruit with a visible surface condition that must move to a different grade;
- fruit that requires manual review or temporary hold;
- fruit assessed against an internal-quality target under an agreed reference method; or
- fruit that is outside the test scope and should not be used to judge the trial.
These questions need different evidence: a size rule can use a measurement; a color boundary can use agreed images and assessor notes; an internal condition may require cutting, measurement, laboratory work, or another fruit-specific method. The label should name the method and the relevant grade rule, because a condition acceptable in one outlet may require downgrading in another.
Define What the Reference Label Must Support
For every category, agree whether the label supports a measurable fact, such as a weight band, or an agreed reference assessment made under stated conditions. The distinction matters for appearance, maturity, and commercial acceptability.
A useful label record normally includes:
| Record item | Why it matters |
|---|---|
| Unique sample ID | Keeps the fruit, images, reference assessment, machine result, and review notes connected. |
| Fruit type, variety, lot, or group | Identifies the relevant fruit basis and variation. |
| Agreed assessment category | States the grade, condition, or decision being reviewed. |
| Reference method | Identifies how the assessment was made and by whom. |
| Assessment time and condition | Records factors that may change relevance. |
| Intended route | Connects the label to the downstream commercial or handling decision. |
| Recheck status | Shows whether the initial assessment was confirmed, changed, or remains disputed. |
A small trial may use a numbered tag, photo set, and shared table; a larger validation may need a formal register. In both cases, the ID must trace from fruit to reference assessment and observed result.
Keep the Reference Method Separate From the Sorting Result
Reference assessment and equipment result should be recorded as separate fields. If they are blended into one label after the trial, it becomes difficult to identify whether a disagreement comes from the reference method, fruit presentation, equipment configuration, routing logic, or the review process.
For a visible external condition, use labeled samples, agreed photographs, manual inspection records, and a stated viewing method as appropriate. The record should not imply that every dark area, scar, cavity, or color variation has the same commercial meaning.
For internal-quality work, the method may involve cutting, an agreed measurement, laboratory analysis, or another documented process. Link the result to the same fruit where possible. If it is based on a sample group, state that limitation before treating it as item-by-item evidence.
Keeping the two records separate is also important for configuration. A machine result may be correct according to the rules currently loaded, while the agreed grade boundary or reference category still needs adjustment. Conversely, an assessment label may be clear while the equipment requires further work. The trial should be able to show which question is being discussed.
Use Clear Categories, Including an Unassessed Category
Ambiguous labels create ambiguous trial results. Categories should be written in a way that a trained assessor can apply consistently under the agreed conditions. Labels such as “acceptable,” “poor,” or “bad” usually need more explanation before they can support a validation discussion.
Include an unassessed or outside-scope category where needed. Fruit may be damaged, outside test conditions, untraceable, or unsuitable for reference review; forcing it into a normal grade distorts the findings. A practical set can include normal commercial, clearly non-conforming, borderline, manual-review, and unassessed fruit.
Record Borderline and Disputed Samples Openly
Borderline fruit often carries more value in a trial than the obvious examples. It is where grade rules, assessor judgment, fruit presentation, and machine configuration are most likely to meet. A trial that uses only obvious premium and obvious reject fruit may look clean while avoiding the daily decisions that affect pack-out and downgrades.
Each borderline sample should retain its ID, evidence, assessor decision, and reason. If the team cannot agree, label it disputed rather than assigning it to the result that favors a demonstration. The disagreement may reveal an incomplete commercial rule, weak reference method, or a need for a different route.
A disputed sample does not make a trial unsuccessful. It makes an uncertainty visible. That can be more useful than an impressive headline percentage with unclear evidence behind it.
Link Labels to the Fruit Condition and Trial Context
For each sample, record the variety, lot, handling and presentation conditions relevant to the agreed reference method. This may include washing, drying, wax, moisture, storage history, fruit orientation, or assessment timing. If the condition changes before review, record the change so the team can judge whether the result still applies to the intended operating condition.
Water, wax, shadows, contamination, orientation, and presentation can change visible appearance; timing, temperature changes, and handling can also affect internal-quality comparisons. A reference label should not imply that an assessment in one state automatically applies to another.
Use Reference Labels in the Trial Review
The final review should compare the reference assessment, the equipment decision, the actual discharge route, and the downstream outcome. It should not reduce the trial to one overall accuracy figure.
| Review question | Evidence to examine |
|---|---|
| Was the reference assessment supported by the agreed method? | Label record, assessor notes, images, measurements, or internal reference record. |
| Did the sorting system make the agreed decision? | Machine result and configuration record. |
| Did the fruit reach the intended route? | Outlet observation, route log, and receiving record. |
| What happens when results differ? | Misclassification type, disputed-label status, and recheck notes. |
| Can the result support an acceptance discussion? | Sample coverage, test condition, remaining exclusions, and agreed responsibilities. |
This review separates different kinds of issue. A false accept may send fruit with an agreed non-conforming condition to a route where it should not go. A false reject may unnecessarily remove acceptable fruit from a premium route. A cross-grade result may send fruit to the wrong saleable category. An unassessed sample should not be treated as proof of either outcome.
Also review the consequence: a result may change premium pack-out, create manual handling, fill the wrong accumulation area, or send fruit to a route without packaging capacity.
Set a Process for Rechecks and Disagreements
Rechecks need a defined process before the trial begins, particularly where assessors may disagree or the reference evidence is destructive. Decide who may reassess a sample, whether photographs or measurements are required, how the original label is retained, and whether the revised result replaces or supplements the initial record.
For uncertain labels or disputed results, agree who can request a recheck, which records need review, when reassessment is no longer possible, and who makes the final reference decision. This prevents uncertain samples from being quietly assigned to the category that makes the demonstration appear stronger.
Where a disputed condition remains unresolved, record it as an exclusion or a future validation question. It should not become an implied capability claim, a final grade rule, or an acceptance condition without further evidence.
Carry the Label Process Into Acceptance Planning
Reference labels can support sample testing, configuration review, FAT, SAT, and troubleshooting when conditions and methods are documented consistently. For acceptance, agree the categories, sample basis, method, operating condition, route evidence, and treatment of exclusions. Separate configured decisions from targets still needing evidence; an untested condition must not become an unconditional performance commitment.
If varieties, grade definitions, presentation conditions, or market routes change, older labels show the earlier decision basis but do not replace new validation.
Frequently Asked Questions
Does every fruit in a trial need an individual reference label?
Not always. Individual IDs are valuable for disputed, borderline, or internal-quality samples. Group records can suit a defined lot-level question if their limitation is documented and the result is not treated as item-by-item proof.
Who should make the reference assessment?
The assessor should understand the agreed grade rule and method. This may involve packhouse quality staff, a buyer representative, technical specialist, laboratory, or more than one reviewer. State the role and method.
Can photos replace physical sample records?
Photos can support visible-condition records when linked to the ID and taken under an agreed approach. They may not capture every angle, presentation state, or internal condition, so use them as part of the method.
What if the reference label and machine result disagree?
First verify the label, the reference method, the fruit identity, and the operating condition. Then review the equipment decision and actual route separately. The disagreement may be a configuration issue, a presentation issue, a disputed reference assessment, or a condition outside the agreed scope. The record should show which explanation remains supported.
Get a Reference-Label & Trial-Validation Checklist
If you are planning a fruit sorting trial, share the fruit type and variety, target grade or condition, reference method, sample availability, current grade rules, and intended route for each result.
SORTTURNKEY can help organize the initial validation discussion around reference labels, sample records, assessor roles, route consequences, and the conditions that need further evidence before configuration.
Request a Reference-Label & Trial-Validation Checklist through the Contact Us page.