// Key takeaways
- Confirm the commodity, destination and current approved marking specification first.
- PUC and packhouse codes are distinct data fields and must not be substituted for one another.
- A camera verifies the visible mark against run data; it does not replace PPECB inspection.
Which fields must your export carton carry?
The PPECB product-inspection page confirms that marking is part of the export inspection scope. It does not supply a universal label layout for all fruit, seasons and destinations. Obtain the current applicable standard, market protocol and approved customer artwork from your export-quality process before creating a camera recipe.
A production-unit code (PUC) identifies the relevant production unit; a packhouse code refers to the packing operation. Store the actual approved codes as controlled master data. Their presence, wording and placement on the carton must follow the specification applicable to that shipment.
What should the line verify?
| Candidate field, where required | Inspection check | Expected data source |
|---|---|---|
| PUC / production-unit identifier | Read the exact value and compare it with the active lot | Approved grower/production-unit record for this run |
| Packhouse identifier | Read and match the approved packing-site code | Controlled packhouse master record |
| Commodity, variety, class, size or count | Check applicable fields against the selected product | Approved label and packing order |
| Batch and date information | Read, match and log the required strings | Production order and traceability record |
| Carton barcode | Decode, check expected data and associate with the carton | Approved product and logistics data |
Where do carton-marking errors enter a run?
A label reel or printer job can remain from the previous grower, product or destination. A correctly printed identifier can therefore be wrong for the carton. Changeover should clear old materials, select the new controlled recipe, synchronise the coder and inspection reference, and hold the line until the first-off mark is accepted.
Include a readable old-lot label in the challenge set. That tests the content comparison, not merely the ability to read a code. Log the recipe revision and run identifier with results so a later review can establish what the system was checking.
What does the camera leave to other controls?
A visible carton check does not inspect fruit quality, phytosanitary condition or actual origin. It cannot prove that incorrect upstream master data is true. It also does not issue an export approval. Keep the official inspection and documentary controls, and test packaging scuffing or damage occurring after the camera separately.
The fruit-carton marking station covers imaging and rejection. Carton and pallet verification covers the later logistics-data relationship.
Follow the data from the packing order to the carton
The most useful carton check starts with the source of each expected field. A packhouse identifier may remain stable while the production unit, variety, class, destination or run changes. Record which fields are fixed master data, which come from the packing order and which are generated during production. If everything is typed into an unrestricted camera recipe, a clear carton label can pass against a manually copied mistake.
PPECB’s public inspection description includes quality, phytosanitary and marking requirements. It does not publish a single universal camera recipe for all commodities and export markets. (PPECB: Product Inspection Services) Build the field list from the applicable approved specification and retain that document’s revision with the recipe. Hortgro’s South African deciduous-fruit context reinforces why a packhouse may handle distinct product and market requirements rather than one generic “fruit export” label. (Hortgro: South African deciduous fruit industry organisation)
Use an explicit release handshake between the production record, printer job and camera reference. Display the active run and key expected fields in a form the operator can compare with the order. A barcode-based job selection can reduce transcription, but it still needs validation that the selected job belongs to the physical fruit being packed. The camera can compare records; it cannot establish origin by looking at the outside of the carton.
A hypothetical old-run label challenge
Hypothetical example: packing changes from Run A to Run B, with a different approved production-unit identifier. An old Run A label remains in the applicator. Its text is sharp and its barcode decodes, but the current expected record is Run B. The carton must be held for the mismatch. This tests content verification; a simple code-present or code-readable check would miss the error. The example does not prescribe any real PUC format or commodity marking requirement.
| Test condition | What it proves | Expected containment |
|---|---|---|
| Old-run label on a current carton | The reference is compared, not merely read | Hold the carton and investigate remaining old labels |
| Correct run, wrong required field | Each mandatory field contributes to release | Hold the mismatch under its specific reason |
| Correct print partly hidden by a strap | The final mark is accessible in the inspection view | Incomplete inspection until visibility is resolved |
| Camera reference not acknowledged | A run cannot start with stale expected data | Hold start or affected product |
| Carton accepted, pallet label inconsistent | Aggregation is checked after carton release | Hold the pallet relationship for reconciliation |
The inspection point should see the mark as it will travel. A loose label may lift after the camera, a strap may cover a required field, or pallet handling may scuff the print. Decide which of those risks is controlled by online imaging and which by packaging checks later in dispatch. If the camera sits before label application, it checks printed labels rather than proving that the correct label was applied to the correct carton.
GS1 General Specifications section 3 defines data structures where GS1 identifiers are used. Keep that logistics syntax separate from commodity-specific marking rules: a valid GS1 symbol does not by itself satisfy every export requirement. Similarly, reading a symbol is different from verifying its print grade with the appropriate verification method. (GS1 General Specifications, Release 26.0)
Make rework and restart traceable
When cartons are relabelled, retain the relationship between the original hold, the corrected label and the final released carton. Removing an error from the camera screen must not remove the reason it was held. Define whether the reworked carton passes through the same station again or follows a controlled offline release route. Record the responsible person and the evidence used to resolve the mismatch without exposing personal production records publicly.
On restart, reconcile cartons between the label applicator, camera and reject. A result belongs to a physical carton, not merely to the latest image. Test a stopped carton with another immediately behind it and verify that the held item is the intended one. If the tracking state cannot be reconstructed, the plant’s procedure should identify the affected group for controlled review rather than assuming the queue is clear.
For image quality, corrugated texture, pale ink and variable label position can matter more than nominal camera resolution. Locate the carton or label before reading the fields, and test dark and light print examples within the agreed printing limits. The HIKROBOT catalogue lists reader and vision-tool capabilities, but those specifications do not establish a packhouse-specific read rate. (HIKROBOT Smart Product Catalog, 2025 Q4)
The acceptance report should identify the commodity and packaging variants tested, the data fields checked, the wrong-value challenges and the physical reject outcomes. Keep no-read, mismatch and incorrect removal counts separate. NIST’s guidance on proportions is relevant to interpreting finite trials, especially when a test happens to contain no failures. (NIST e-Handbook, 7.2.4.1 Confidence intervals for a proportion) Use the associated fruit-carton solution page for station layout and sample inputs. Official export inspection, documentary controls and shipment release remain separate responsibilities; this article describes production verification that supports those controls.
Turn the marking rules into a carton inspection scope
Use the approved packing data and representative cartons to define the reading task, the data comparison and what happens to a carton that fails.
Fruit carton marking inspection
Review camera and code-reader placement, expected run data and reject handling on the carton conveyor.
Carton and pallet verification
See how carton labels and pallet identities are checked against controlled order or warehouse data.
Frequently asked questions
Is there one PPECB label template for all exports?
Can a PUC check be only a barcode scan?
Does a readable carton code prove the correct grower?
Can automated carton inspection replace PPECB?
Frank Guo is addaScan’s technical lead and writes the technical articles on this site. More about the team → LinkedIn
Sources
- PPECB Product Inspection Services: export standards and marking requirements (Official source; checked 2026-10-26)
- Hortgro: South African deciduous fruit industry organisation (local industry context; read 2026-10-27)
- GS1 General Specifications, Release 26.0 (sections 3 and 5: data and symbol requirements; read 2026-10-27)
- HIKROBOT Smart Product Catalog, 2025 Q4 (vendor catalogue; archived text checked 2026-10-27)
- NIST e-Handbook, 7.2.4.1 Confidence intervals for a proportion (binomial interval methods; read 2026-10-27)