Fruit-carton markings and packing records
Fruit-carton marking verification should compare readable fields with the active packing record. Hold on business mismatch, not image confidence.
Carton marking inspection is a check of every fruit carton leaving the packline that the end marking, such as the producer and packhouse codes, class, count and date code, is present, legible, upright and right for the fruit being packed, and that the carton barcode reads and agrees. addaScan builds it per packline on a HIKROBOT code reader and camera with a carton pusher. An indicative installed budget per line for a typical application is R 210 000 – R 940 000 excluding VAT, including travel.
Fruit export is packed carton by carton. The Citrus Growers’ Association reports that Southern African citrus growers packed 203.4 million 15 kg cartons for global markets in 2025 (CGA, 2025), a figure that includes growers in neighbouring countries. In a citrus packhouse each carton is marked at the end by an inkjet printer, a stamp or a carton labelling machine, and that marking is what ties the carton back to the orchard and the packhouse.
For citrus sold in South Africa, R963 requires the particulars to be marked “clearly, indelibly, legibly and not untidy, upside down or askew”, grouped on one side of the carton (R963, 2012). They include the class, size code, count, the producer or packhouse code and a date code on at least 90% of the containers (R963). For export, PPECB inspection covers quality, phytosanitary aspects and marking requirements (PPECB); what each market needs on the carton comes from the export standards and your exporter, and the station is set up with that list.
The rules set the outcome, not the method. A camera at the carton end checks every carton before it is palletised, so a printer fault or a wrong PUC code is found on the line rather than at inspection.
To define the expected fields before specifying the station, use the PPECB export carton marking guide. It follows PUC, packhouse and run data from the packing order to the printed carton, including mismatches and changeovers.
| What the line reports | What a station checks | Where it stops |
|---|---|---|
| “The printer ran dry and we stacked half a pallet without codes” | Marking present on every carton end, every field in its place, read by OCR. | Marking on a carton side the camera does not face is not seen; cartons must arrive with the marked end towards it. |
| “Wrong PUC on the cartons after a bin change” | The PUC code, packhouse code, class and count read from the carton are compared with the run data from the packhouse system or the line recipe. | The station knows only what the system tells it; it cannot tell from the carton which orchard the fruit came from. |
| “Stamps on skew or upside down” | Marking position and orientation on the carton end, so skewed and inverted marking is rejected. | Limits for position and angle are set from your marking specification; the station does not set them. |
| “The carton barcode doesn’t scan at the cold store” | The carton barcode read on every carton and its content compared with the run. | Reading a code is not grading it; GS1 print quality is measured with a verifier. |
| “Faint print on wet or dusty cartons” | Weak or broken characters found by OCR confidence and, in the fuller station, by deep-learning OCR trained on your cartons. | Print too faint for a person to read is rejected, not guessed. |
Packed cartons travel from the packing tables or the lidder towards the palletising area with the marked end facing one side. A code reader and a camera there read the marking on every carton. In the fuller station the other end is read too, and the result is compared with the run data from the packhouse system.
Swipe sideways to see the whole drawing →
| Component | HIKROBOT example | Role in this station |
|---|---|---|
| Code reader | Low: HIKROBOT ID5000 code reader MV-ID5060M-12S-WBN. High: two ID5000X code readers MV-ID5050XM-12M-RBN with deep-learning decoding, one per carton end | Reads the carton barcode |
| Marking camera | Low: HIKROBOT SC3000 vision sensor MV-SC3050M with OCR and character comparison. High: two MV-CS200-10GM 20 MP mono area scan cameras with MVL-HF2524M-10MP lenses | Reads the PUC and packhouse codes, class, count and date code |
| Processing (high) | HIKROBOT VC3000 vision controller (MV-VC3302P) running HIKROBOT VisionMaster with the deep-learning licence | Runs deep-learning OCR for both ends and compares the result with the run data |
| Lighting | HIKROBOT MV-LLDS-H-600-30-W bar light, one per carton end | Even light across the marking on brown or white board |
| Trigger | HIKROBOT MV-PE photoelectric sensor | Fires the reader and camera once per carton |
| Reject | Not HIKROBOT: carton pusher with confirmation sensor (high); low case signals the existing reject or stops the conveyor | Moves failed cartons to rework before palletising |
| Packhouse data link (high) | Local supply: interface to the packhouse system or line PLC | Gives the station the PUC, class, count and date expected for the run |
| Tool or software | What it does here |
|---|---|
| HIKROBOT IDMVS (code reader client) | Sets up the reader and outputs the decoded carton barcode. |
| HIKROBOT SCMVS OCR and character comparison (low) | Reads each marking field and compares it with the values set for the run. |
| VisionMaster deep-learning OCR (high) | Reads inkjet and stamped characters on corrugated board where the print is uneven or low in contrast. |
| Position and orientation tools | Find the marking block and reject it when skewed, misplaced or upside down. |
| Run data and log | Expected values per run from the packhouse system or recipe; a log of rejected cartons with the reason and image. |
Reader, camera and light on a rigid side mount beside the carton conveyor, clear of carton corners and forklifts; covers against wet cartons and packhouse dust.
A bar light along the carton end, angled so wet or waxed board does not glare; a shroud keeps sunlight from open doors off the marking.
One trigger per carton; cartons must arrive with the marked end on the camera side and with a gap, and each carton is tracked to the pusher.
Result and reject reason over PROFINET or EtherNet/IP (listed for these HIKROBOT readers) or digital I/O; run data from the packhouse system in the fuller station.
Carton pusher with a confirmation sensor and a full-table signal; if the vision system is not ready, cartons are held, not passed unchecked.
A recipe per commodity, carton type and market marking; a screen at the line shows the last rejected carton with the field that failed.
Indicative budget for a typical application of this kind. Every project involves custom requirements (your products, line speed, layout, PLC and acceptance criteria), so the final price differs from case to case. A firm price follows the sample test.
Figures are per packline. The low figure is a code reader and a vision sensor on one carton end, with the marking compared with values set for the run and an existing reject signalled. The high figure reads both carton ends with deep-learning OCR on a controller, takes the run data from the packhouse system and adds a carton pusher with confirmation. Travel is included at an estimate for a packhouse outside Gauteng, which the general cost guide leaves out; a packhouse with several lines shares travel and some engineering. Changes to your conveyor, the packhouse system vendor’s interface work, utilities and VAT are excluded.
Hardware is priced from public list prices converted at the ECB reference rate, plus freight, clearance and margin; engineering at R 900 – R 1 300 per hour. How these ranges are built →
Swipe the table sideways to see the high figure and the basis →
| Cost block | Low | High | Basis |
|---|---|---|---|
| Vision hardware (landed, with margin) | R 95 600 | R 288 800 | Low: one ID5000 code reader, one SC3000 vision sensor, bar light. High: two ID5000X deep-learning code readers, two 20 MP mono cameras with lenses, two bar lights, VC3000 controller |
| Software licence | R 0 | R 120 000 | Low: none, the reader and vision sensor software is free. High: HIKROBOT VisionMaster licence with deep learning |
| Local materials and reject | R 20 000 | R 100 000 | Low: mounts over the line, cabling. High: mounts, carton pusher with confirmation, panel |
| Engineering (recipes, PLC link) | R 45 000 | R 208 000 | 50 h at R 900 to 160 h at R 1 300 |
| Installation, FAT, SAT, training, documents | R 21 600 | R 83 200 | 24 h to 64 h |
| Project management | R 9 000 | R 41 600 | 10 h to 32 h |
| Travel | R 20 000 | R 50 000 | Estimated flights and accommodation to a packhouse region |
| Total (rounded) | R 210 000 | R 940 000 | Per line. Low: one carton end. High: both ends, deep learning, data link |
The technologies behind this station, how much each matters here, and where each is explained in depth.
| Technology | Role | Why it matters in this scenario | Read more |
|---|---|---|---|
| OCR/OCV of printed codes | Core | The PUC and packhouse codes, class, count and date code are text on the carton end; each field is read and checked. | OCR/OCV inspection |
| Fixed code reader (HIKROBOT ID series) | Core | The carton barcode is read on every carton, on board that may be wet, waxed or dusty. | Choosing a code reader |
| Code content matched to PLC, MES or WMS data | Core | A legible but wrong PUC or count is the costly fault; the reading is compared with the run data. | Carton and pallet verification |
| Deep learning (classification, segmentation, DL OCR) | Supporting | Inkjet and stamp print on corrugated board is uneven; deep-learning OCR trained on your cartons reads it more consistently. | Rule-based vs deep learning |
| Lighting technique | Supporting | Brown board, white board and waxed cartons reflect differently; the bar light is set per carton type. | Machine vision lighting |
| PLC link and reject with confirmation | Supporting | Full cartons are heavy; the pusher needs tracking, confirmation and a rework table. | PLC and reject integration |
Every station is accepted on your products, against numbers agreed in writing before the build. How we deliver →
Cartons of each type and marking method from a normal packing day, good and faulty (missing field, faint print, skewed or inverted marking, wrong code, damaged barcode), imaged with the proposed reader, camera and light.
Before delivery, marked challenge cartons must each be rejected for the right field, and correctly marked cartons of each recipe must pass.
On your packline in season, the challenge cartons are run again at packing speed, and false rejects and escapes are counted over an agreed run of production.
It is a check of every packed carton, before palletising, that the marking on the carton end is present, legible, upright and correct for the fruit being packed: producer and packhouse codes, class, count, date code and any other field your marking specification lists. The carton barcode is read at the same time. Cartons that fail go to a rework table.
No. PPECB inspects export consignments for quality, phytosanitary aspects and marking requirements, and only its inspectors approve a consignment. The station is a packhouse control that checks every carton’s marking on the line, so marking faults are found and fixed before the inspector sees the pallet. It does not look at the fruit.
An indicative installed budget is R 210 000 – R 940 000 per packline excluding VAT, at 2 October 2026 prices, including travel to the packhouse. The low figure reads one carton end with a code reader and a vision sensor; the high figure reads both ends with deep-learning OCR, takes run data from the packhouse system and adds a carton pusher. A firm price follows a sample test on your cartons. This is a typical application: every project involves custom requirements, so the final price depends on your products, line and acceptance criteria.
It checks the PUC code against what the packhouse system or the line recipe says is being packed. If the system knows which bins are on the line, a carton with a different PUC is rejected. The camera cannot tell from the carton where the fruit came from, so the check is only as good as the run data it receives.
It reads the barcode on every carton and checks its content, so unreadable and wrong codes are rejected. GS1 sets a minimum print quality for distribution barcodes, which is measured with a barcode verifier, usually on samples. The line reader and the verifier do different jobs and are used together.
Describe the line, the products and the fault. We test on your good and faulty samples and reply with a feasibility note and a budget.