Wine front and back label verification
Wine-bottle label verification needs to match the front and back labels to the active bottle recipe. A small text difference changes the required identity.
Wine bottle label inspection is a camera check of every bottle after the labeller: front and back labels are present, straight, at the right height and from the right run, the capsule is on, and the certification seal and any printed code are where they should be. addaScan builds it on four to six HIKROBOT cameras on one controller with a bottle reject. An indicative installed budget for a typical application is R 350 000 – R 1 170 000 excluding VAT.
South Africa sells a large volume of wine at home: SAWIS reports total domestic sales of 407 514 165 litres in the twelve months to February 2026 (SAWIS, 2026). Much of it goes through a wine bottling line where a bottle labelling machine applies a front and a back label, often a neck label, and a capsule goes on over the closure. Every label run, vintage and cultivar change is a chance for the wrong reel or a label that lands crooked.
The Liquor Products Act regulations set the marking. A lot identification is any mark identifying the lot, preceded by the letter “L”, and a filling date is shown as “Filled on” with the date (Liquor Products regulations). The same regulations say a lot identification is not required for a wine with a certification seal, and the filling date may be omitted on a glass container (reg 33(2)). The certification seals now carry the Wine Certification Authority name and a QR code (SAWIS, 2025). So what a bottle must carry depends on the wine and the scheme. The station is set up per product: it checks for the seal, a lot code or a filling date only where your label specification calls for them.
The regulations set what the label says, not how a producer checks it. Cameras after the labeller are one way to see every bottle rather than a few per pallet.
| What the line reports | What a station checks | Where it stops |
|---|---|---|
| “Back labels missing on some bottles” | Front, back and neck label present on every bottle, from cameras that together see the whole label band. | A label stuck inside the case or on another bottle is not the station’s concern; it sees only the bottle in front of it. |
| “The labels are going on skew, or too high” | Label alignment: angle, height and gap between front and back label, compared with limits set on the master bottle. | Limits are set from your label specification and samples; the station does not decide what is acceptable. |
| “Last vintage’s back label on this year’s wine” | The label is compared with the master for this run (pattern and text), so a label from another product or vintage is rejected. | Two labels that differ only in very small print need a camera close enough to read it; this is tested on samples. |
| “Flags and wrinkles on the labels” | Label edges that lift, folds and creases that change the label outline or the light reflected from it. | Small wrinkles on embossed or foil labels can look like the label’s own texture; these are judged on samples. |
| “No capsule, or no seal on the capsule” | Capsule present and seated over the neck, and the certification seal present where your bottle carries it, from a top or neck view. | Whether the capsule is crimped or spun tight is not measured; the station sees its outline only. |
Bottles leave the labeller in single file. Four cameras around the conveyor see the full label band; in the fuller station a fifth looks down on the capsule and a sixth reads the printed code. One controller combines the views per bottle and signals the reject.
Swipe sideways to see the whole drawing →
| Component | HIKROBOT example | Role in this station |
|---|---|---|
| Label cameras | Low: four HIKROBOT MV-CS060-10GM 6 MP mono area scan cameras with MVL-HF1228M-6MPE lenses. High: MV-CS200-10GC 20 MP colour cameras with MVL-HF2524M-10MP lenses | See the whole label band around the bottle; colour shows the wrong label by colour as well as by print |
| Capsule and code cameras (high) | Two further MV-CS200-10GC colour cameras: one over the capsule, one on the printed code | Capsule and certification seal from above; lot code or filling date where printed |
| Processing | HIKROBOT VC3000 vision controller (MV-VC3302P low, MV-VC3704H high) running HIKROBOT VisionMaster | Combines all views per bottle and sends one result to the PLC |
| Lighting | Four HIKROBOT MV-LLDS-H-600-30-W bar lights for the labels; high adds a HIKROBOT MV-LRDS-H-160-00-W ring light and a HIKROBOT MV-LGSS dome light, supplied by addaScan | Even light on curved glass and labels; the dome light tames reflective capsules |
| Trigger | HIKROBOT MV-PE photoelectric sensor | Fires every camera for the same bottle |
| Reject | Not HIKROBOT: bottle pusher with confirmation sensor and reject table or bin (high); low case signals the labeller’s or line’s existing reject | Removes failed bottles without stopping the line |
| Mounting | Local supply: camera mounts and light shroud (low); mounting frame and light tunnel, panel, screen and UPS (high) | Keeps camera positions fixed through changeovers |
| Tool or software | What it does here |
|---|---|
| HIKROBOT VisionMaster pattern and position tools | Find each label and measure its angle, height and the gap to the next label. |
| VisionMaster Image Stitch | Join the side views into one picture of the label band, so a label seen half in two cameras is judged once. |
| VisionMaster OCV and OCR | Compare printed label text and the lot code or filling date with the expected text for the run. |
| VisionMaster colour tools (high) | Confirm label and capsule colour for the product, which separates look-alike labels from different wines. |
| Recipe per product | One recipe per wine, label set and bottle shape, with its own list of what must be present (seal, code, neck label). |
Cameras on a rigid frame around the conveyor after the labeller, clear of the labeller’s moving parts and easy to clean around; a light tunnel in the fuller station.
Bar lights placed so the curved glass and glossy or foil labels do not throw glare into the cameras; a shroud blocks daylight from the bottling hall.
One trigger for all cameras; bottles must run in single file with a gap, and the PLC or controller tracks each bottle to the reject.
Result, reject reason and recipe number over PROFINET or EtherNet/IP where the chosen HIKROBOT device lists it, or digital I/O from the controller on simple stations.
Pusher with a confirmation sensor and a full-table signal; if the vision system is not ready, bottles are rejected or the line is held, not passed unchecked.
Recipe chosen with the label run at changeover; a screen shows the last rejects with the reason and the image from the camera that failed it.
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 four mono cameras on one controller checking that the labels are present, straight and the right ones, signalling an existing reject. The high figure uses six colour cameras that add the capsule and seal from above and a reading of the printed code, with a bottle pusher, confirmation and bin. Both include travel to a Western Cape winery or bottler, which the general cost guide leaves out. Deep-learning label classification is not included. Changes to your conveyor or labeller, bottle spacing hardware, 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 116 200 | R 355 200 | Low: four 6 MP mono cameras with lenses, four bar lights, VC3000 controller. High: six 20 MP colour cameras with lenses, bar lights, dome and ring light, larger VC3000 controller |
| Software licence | R 33 800 | R 50 100 | HIKROBOT VisionMaster licence for four cameras; high case with the tier for more cameras |
| Local materials and reject | R 50 000 | R 160 000 | Low: four mounts, light shroud, network switch, cabling, trigger. High: mounting frame and light tunnel, bottle pusher with confirmation and bin, panel, screen and UPS |
| Engineering (recipes, PLC link) | R 72 000 | R 286 000 | 80 h at R 900 to 220 h at R 1 300 |
| Installation, FAT, SAT, training, documents | R 36 000 | R 135 200 | 40 h to 104 h |
| Project management | R 18 000 | R 62 400 | 20 h to 48 h |
| Travel | R 25 000 | R 60 000 | Estimated flights and accommodation for a Western Cape site |
| Total (rounded) | R 350 000 | R 1 170 000 | Low: four mono label views. High: six colour views and 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 |
|---|---|---|---|
| Area scan cameras with a vision controller | Core | A label wraps around a round bottle; several fixed cameras on one controller see the whole band without turning the bottle. | Camera and lens selection |
| Label inspection | Core | Label present, aligned and correct for the run is the main job of this station. | Label inspection |
| HIKROBOT VisionMaster tools | Supporting | Image Stitch joins the side views, and pattern and position tools measure each label. | VisionMaster algorithm tools |
| OCR/OCV of printed codes | Supporting | A lot code or filling date, where printed, is read and compared with the run. | OCR/OCV inspection |
| Colour identification | Supporting | Labels of different wines often share a layout and differ in colour; capsules differ by colour too. | Surface and colour inspection |
| Lighting technique | Supporting | Curved glass, foil and embossed labels reflect; light placement decides what the cameras can see. | Machine vision lighting |
Every station is accepted on your products, against numbers agreed in writing before the build. How we deliver →
Bottles of your main products with good labels and with each fault (missing, skewed, wrong vintage, wrinkled, no capsule, no seal), imaged with the proposed cameras and lights; the result says what is reliably caught.
Before delivery, marked challenge bottles must each be rejected for the right reason, and correctly labelled bottles of each recipe must pass.
On your bottling line at production speed, the challenge bottles are run again, and false rejects and escapes are counted over an agreed run of production.
It checks every bottle after the labeller: front, back and neck labels present, at the right height and angle, from the right run and without large wrinkles or lifted edges; the capsule on; the certification seal where your bottle carries one; and a lot code or filling date where you print one. Bottles that fail are pushed out of the line.
Not always. Under the Liquor Products Act regulations, a lot identification starts with the letter “L”, but it is not required for a wine with a certification seal, and the filling date may be left off a glass container. The station therefore checks per product, against what your label specification says that product carries. Confirm the requirement for each product with your compliance adviser.
An indicative installed budget is R 350 000 – R 1 170 000 excluding VAT, at 2 October 2026 prices, including travel to a Western Cape site. The low figure is four mono cameras checking the labels and signalling your existing reject; the high figure is six colour cameras that add the capsule, seal and printed code, with a bottle pusher and confirmation. A firm price follows a sample test on your bottles. This is a typical application: every project involves custom requirements, so the final price depends on your products, line and acceptance criteria.
Usually, yes, with the cameras placed for that bottle. Height, angle and the gap between labels are measured against limits set on a correct bottle, so each bottle shape gets its own recipe and camera positions are checked at changeover. Strongly tapered shoulders, embossed glass and foil labels are included in the sample test, because they change what the cameras see.
The main check is that the seal is there and in the expected place. Reading the QR code as well is possible only if the code faces a camera and is large enough in the image, which depends on where the seal sits on your bottle and capsule. That is decided in the sample test, not assumed.
No. Four fixed side cameras together see the whole label band while the bottle passes, so no turning mechanism is needed. A rotating station with a line scan camera is an alternative, but it needs bottle handling that changes your conveyor, and is not part of the budget on this page.
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.