Shrink-wrap code reading
Test glare and film distortion before choosing OCR or expected-code comparison.
Multipack inspection is a camera count of every tray or shrink-wrapped pack after the packer: it confirms each position holds a bottle or can, that caps are on, and that the pack holds the product the recipe expects. addaScan builds it on a HIKROBOT vision sensor for a single top view, or two HIKROBOT cameras on a vision controller with a pusher reject. An indicative installed budget for a typical application is R 80 000 – R 660 000 excluding VAT.
Soft drinks, water and beer leave South African plants in shrink-wrapped packs and trays as well as in crates. Coca-Cola Beverages South Africa alone runs 11 plants across six provinces (CCBA), and Nampak makes beverage cans and ends at plants in Johannesburg and Pretoria for the Southern African beverage industry (Nampak). On most of these lines a tray packer or case packer groups the bottles or cans, and a shrink tunnel wraps the pack.
A pack with a missing bottle or can is short against what the pack says it holds. The Legal Metrology Regulations require prepackages to comply with SANS 289 (Legal Metrology Regulations, 2018), and FSSC 22000 asks food sites on the scheme to control quantity by unit, weight and volume (FSSC 22000 v6, 2.5.9(b)). Neither tells you how to check. A count of every pack at the end of the line is one way to support the declared quantity, pack by pack, instead of finding short packs in the warehouse or at the customer.
The camera can only count what it can see. Through clear shrink wrap it sees bottle caps and can ends from above; through printed or opaque film it sees nothing, so on those packs the count is done before the shrink tunnel, on the open tray.
| What the line reports | What a station checks | Where it stops |
|---|---|---|
| “The customer found a pack with a can missing” | Count of every position in the pack from above: each bottle cap or can end where the pack layout says one should be. | A can hidden under printed film, or lying flat under the others, is not counted from above. |
| “The tray packer left a hole after a jam” | The open tray straight after the tray packer, before the shrink tunnel: every position filled, no bottle lying across the tray. | A bottle lying across the tray is rejected only once that fault has been taught on samples. |
| “A bottle without a cap went out in a shrink-wrapped pack” | Cap present on each bottle, seen from above through clear film. | Cap tightness and the tamper band are not checked; that is the job of the capper and a cap inspection at the filler. |
| “Wrong flavour in the pack after the changeover” | The cap colour, can end or top print seen from above, matched to the product recipe the PLC has selected. | Products that look identical from above need the side camera in the two-view configuration; the label is not read through printed film. |
| “Packs coming out of the tunnel torn open” | Large tears and open ends in the film, seen from the top and side views. | Film shrink quality and pack tightness are not measured. |
A single vision sensor looks down on each pack as it passes, or, for the fuller station, a top camera and a side camera feed a vision controller. The result goes to the PLC, which pushes incomplete packs into a reject lane and confirms they left.
Swipe sideways to see the whole drawing →
| Component | HIKROBOT example | Role in this station |
|---|---|---|
| Vision sensor (single view) | HIKROBOT SC3000-series vision sensor with built-in light (budget priced on MV-SC3016C); SC3000X (e.g. MV-SC3013XC) where learning-based count is needed | Counts every position from above and signals the reject |
| Cameras (two views) | 2 × HIKROBOT MV-CS050-10GM 5 MP mono GigE cameras with 25 mm lenses (MVL-HF2524M-10MP) | Top view for the count; side view for product type and film tears |
| Processing (two views) | HIKROBOT VC3000 vision controller (e.g. MV-VC3302P) running HIKROBOT VisionMaster | Runs one recipe per pack format and sends the result to the PLC |
| Lighting | HIKROBOT MV-LRDS-H ring light on the single-view station; on the two-view station a dome light over the pack and an MV-LLDS-H bar light for the side view | Diffuse light that keeps shrink-film glare off the caps and can ends |
| Trigger | HIKROBOT MV-PE photoelectric sensor or the existing pack sensor | Fires the camera when the pack is under it |
| Reject | Not HIKROBOT: pneumatic pusher with confirmation sensor and reject bin (two-view station); the single-view station signals your existing reject | Removes incomplete packs and proves they left |
| Mounting and panel | Local supply: brackets, shrouds, cabling, panel and operator screen | Holds the views fixed and shows rejects to the operator |
| Tool or software | What it does here |
|---|---|
| SCMVS count tools on the vision sensor | Spot, contour and pattern count of caps or can ends in each position of the pack. |
| Learning-based count (SC3000X) or VisionMaster deep-learning object detection | Option for packs where caps overlap or film creases break up the outline; trained on your packs. |
| VisionMaster pattern matching per product | Matches the cap, can end or top print to the product the recipe expects. |
| Recipe per pack format | One recipe per pack format and product, chosen by the PLC at changeover. |
| Result and image log | Reject reason per pack and saved images of rejected packs for the line team. |
Camera or vision sensor on a bracket over the pack conveyor, high enough for the tallest pack; the side camera at a fixed distance from a guide rail so every pack passes at the same depth.
A diffuse light and a shroud over the top view, so the plant lights do not reflect off the shrink wrap. The SC3000X spec sheet lists polarised and diffuser lens caps for the same reason.
One trigger per pack; the PLC tracks each pack from the camera to the pusher so the right pack is removed.
Pass, fail and reject reason over digital I/O, or over PROFINET or EtherNet/IP, which the SC3000X spec sheet lists; on the controller station, I/O or fieldbus to the line PLC.
Pusher with a confirmation sensor and a full-bin signal; if the vision system is not ready, packs are rejected or the line is held, not passed unchecked.
Pack format and product recipes chosen by the PLC at changeover; a screen shows the last rejected packs and the reason.
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.
The low figure is one vision sensor above the pack that counts caps or can ends and signals your existing reject. The high figure adds a side view, a vision controller with product recipes and a new pusher reject with confirmation. Deep-learning counting is not in either figure; if your packs need it, it is priced after the sample test. Changes to your conveyor, 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 40 100 | R 200 000 | Low: one colour vision sensor and a ring light. High: two 5 MP cameras with lenses, a long bar light, a dome light and a VC3000 controller |
| Software licence | R 0 | R 40 100 | Vision sensor software is free; high case with the HIKROBOT VisionMaster standard licence |
| Local materials and reject | R 8 000 | R 110 000 | Low: bracket, cabling, light driver, signal to the existing reject. High: mounts and shrouds, pusher reject with confirmation sensor and bin, panel and screen |
| Engineering (recipes, PLC link) | R 18 000 | R 156 000 | 20 h at R 900 to 120 h at R 1 300 |
| Installation, FAT, SAT, training, documents | R 10 800 | R 83 200 | 12 h to 64 h |
| Project management | R 3 600 | R 31 200 | 4 h to 24 h |
| Travel (Gauteng) | R 1 500 | R 6 000 | Outside Gauteng quoted separately |
| Total (rounded) | R 80 000 | R 660 000 | Low: one vision sensor. High: top and side views with own reject |
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 |
|---|---|---|---|
| Presence detection and counting | Core | Every position in the pack is a presence check; a missing bottle or can is a missing cap or can end in the count. | Presence and counting |
| Vision sensor (HIKROBOT SC1000–SC3000) | Core | One top view with fixed pack formats suits a vision sensor with count tools and no PC. | Choosing a smart camera |
| Area scan cameras with a vision controller | Supporting | The two-view station needs a top and a side camera on one controller to judge product type and film tears. | Camera and lens selection |
| Lighting technique | Supporting | Shrink wrap reflects light like a mirror; a diffuse or dome light and a shroud decide whether the count is stable. | Machine vision lighting |
| Deep learning (classification, segmentation, DL OCR) | Supporting | Learning-based count or object detection copes with overlapping caps and creased film where rule-based count fails. | Rule-based vs deep learning |
| PLC link and reject with confirmation | Supporting | The right pack must be pushed off and the reject confirmed, or a short pack still reaches the pallet. | PLC and reject integration |
Every station is accepted on your products, against numbers agreed in writing before the build. How we deliver →
Good packs of each format plus packs with a missing bottle, a missing cap and the wrong product, imaged with the proposed camera and light, through the film you actually use; the result says what is reliably found and what is not.
Before delivery, a set of marked challenge packs must each be rejected, and good packs of every format must pass.
On your line, at production speed, the challenge packs are run again, and false rejects and escapes are counted over an agreed run of production.
Missing-bottle detection is only reliable when the relevant positions can be observed. Transparent film can still reflect, wrinkle or obscure a closure.
| Decision or event | Check or action | Acceptance boundary |
|---|---|---|
| Expected pattern | Bind row/column occupancy to the recipe | The same count in the wrong arrangement may still be incorrect. |
| Visible cap/top | Check each exposed position | Hidden lower layers need another point or method. |
| Film variation | Challenge normal wrinkles and glare | Do not teach a reflection as a product. |
| Reject confirmation | Prove the intended pack left the main flow | Correct image classification does not prove removal. |
Test adjacent packs and altered spacing as well as missing items. Treat ambiguous views as a defined exception, not an inferred full pack.
Multipack inspection is a camera check of every tray or shrink-wrapped pack after the packer. It counts the bottles or cans by their caps or can ends, checks that caps are on and that the product matches the recipe, and signals a reject for incomplete packs. It sits at the end of the packing line, before the palletiser.
Through clear shrink wrap, yes, when the light is diffuse and the camera is shrouded so the film does not throw glare into the image. Through printed or opaque film the camera cannot see the bottles at all. On those packs the count is done on the open tray straight after the tray packer, before the shrink tunnel.
An indicative installed budget is R 80 000 – R 660 000 excluding VAT, at 2 October 2026 prices. The low figure is one HIKROBOT vision sensor above the pack, signalling your existing reject. The high figure is a top and a side camera on a vision controller with product recipes and a new pusher reject with confirmation. Both include engineering, installation, FAT and SAT in Gauteng. This is a typical application: every project involves custom requirements, so the final price depends on your products, line and acceptance criteria.
Missing bottles, yes: the count in the open tray is checked before shrink wrapping. Underfilled containers are best caught by a fill level check on the single-file line before packing, described on the fill level page; once packed, only clear containers in the outer row show their level.
Before the tunnel, on the open tray, the camera has the clearest view and works with any film. After the tunnel it also catches packs damaged in the tunnel, but it needs clear film. Where the film is printed, the check goes straight after the case packer or tray packer; otherwise the sample test on your film decides.
No. The camera counts bottles and cans and checks caps and product type from outside the pack. Fill level is checked at the filler, and pack weight by a checkweigher. A count from the camera and a weight from the checkweigher answer different questions.
Yes. On a bottle case packer the camera counts caps in each case before it is closed; on can packs it counts can ends. Each pack format gets its own recipe, and the missing bottle and missing can faults are both included in the sample test.
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.