Visible can-end and seam inspection
Camera inspection of a metal can end can check visible edge irregularities, lid condition and agreed external seam features. An external protrusion or depression needs assessment.
Line solution
Test this on your samplesCan inspection: end codes and dented cans
// Short answer
Can inspection is a camera check of every filled can after coding: it reads the date and batch code on the can end, compares it with what the line should be printing, and rejects cans with a missing, smudged or wrong code. The fuller station adds side cameras that look for visible dents in the can body. addaScan builds it on HIKROBOT hardware. An indicative installed budget for a typical application is R 100 000 – R 1 180 000 excluding VAT.
// At a glance
The station in three lines
- Faults caught
- Missing, smudged or wrong codes on the can end; visible dents in the can body (two-station option)
- Configuration
- HIKROBOT vision sensor with OCR over the can end; optional four side cameras on a VC3000, air-jet reject
- Line link
- Digital I/O or fieldbus to the line PLC; expected date and batch sent by the PLC
// Fit & limits
What this station will not do, and what it needs
What it will not do
- Seams and leaks: double seam faults, pinholes and leakers are not visible from outside; they need seam measurement and leak testing.
- Inside the can: fill level, vacuum and contents are not checked.
- Code grading: reading and checking a code is not grading its print quality to a standard.
- Every dent: a shallow dent, or a dent on the seam, may not show in a 2D image; dent detection is proven on your samples before it is promised.
What makes it work
- Samples: good cans, cans with the code faults you actually see, and real dented cans from your line.
- Expected values: the date and batch for each run, sent by the PLC or entered once at changeover.
- Spaced cans: cans separated enough at the side cameras and the air jet to see and reject one at a time.
- A reject route: a guide and bin for blown-off cans, and someone who checks them.
// On South African lines
Can coding and dented cans on South African lines
Cans are made and filled at scale in South Africa. Nampak describes itself as the leading South African maker of food cans and ends (Nampak), and its beverage can plants in Johannesburg and Pretoria supply the Southern African beverage industry (Nampak). Whether the line fills a beverage can or canned food, the code is usually printed or stamped on the can end after seaming, and every can then runs at full speed to the packer.
The rules set the outcome. The canned vegetable regulations require each container to be clearly and legibly marked with a code mark that gives the canning establishment, the product and the date of manufacture (R1532, 2003). The food hygiene regulations prohibit selling canned food in a container that is so dented or damaged that it may contaminate the food, or that bulges, leaks or has become unsealed (R638, 2018). Can coding that fails for a few minutes leaves a run of cans without a traceable code; a camera reading every can end is one way to catch that at once, and to divert the cans rather than rework a pallet.
The regulator’s own list of typical canned-food faults is seam defects, leaking cans and cans with pinholes (NRCS, 2023). Those are not visual checks from outside the can, and this station does not claim them.
What the line reports, and what can inspection checks
| What the line reports | What a station checks | Where it stops |
|---|---|---|
| “The coder stopped and nobody noticed until the next QA check” | Code present on every can end, read by OCR straight after the coder. | A code printed on the wrong place of the end, outside the camera’s field, is reported as missing, not found. |
| “The date is there but half of it is missing” | Each character checked against the expected string (OCV) or by a deep-learning character-defect check trained on your good and bad prints. | Legibility is judged against your accepted samples; it is not a grade to a print-quality standard. |
| “Yesterday’s date was still on the coder after the shift change” | The read date and batch compared with the values the PLC sends for this run. | If the PLC or recipe sends the wrong expected date, the camera checks against the wrong date. |
| “Dented cans keep turning up on the pallet” | Side views around the can body (two-station option), classified by deep learning trained on your dented and good cans. | Small or shallow dents that hardly change the reflection are proven on samples first; a dent on the seam itself is not judged. |
| “We had leakers in the warehouse” | Nothing: a leak, pinhole or poor double seam is not visible from outside the can. | Seam measurement and leak testing stay with your seam inspection and QA routine. |
// The complete solution
Can inspection station: hardware, software and installation
A vision sensor with OCR sits over the conveyor after the coder and reads every can end. For dents, four side cameras around a short light tunnel see the whole can body. Failed cans are blown off by an air jet into a reject guide, and a sensor confirms they left.
[Schematic]Plan view, not to scale
Swipe sideways to see the whole drawing →
[1]: A photoelectric sensor tells the vision sensor a can has arrived.
[2]: Vision sensor with OCR and a dome or ring light reads the code on the can end.
[3]: Four side cameras look for visible dents around the can body (option).
[4]: An air knife on a fast valve blows failed cans into a reject guide.
[5]: Confirms the rejected can left the conveyor.
Hardware
| Component | HIKROBOT example | Role in this station |
|---|---|---|
| Code vision sensor | HIKROBOT SC3000-series vision sensor with OCR, 5 MP mono (budget priced on MV-SC3050M) | Reads the date and batch on every can end |
| Code lighting | HIKROBOT MV-LRDS-H ring light (code only); a dome light for shiny or embossed ends on the full station, such as HIKROBOT’s MV-LBES-H flat dome | Even light on a curved, reflective can end |
| Side cameras (dent option) | 4 × HIKROBOT MV-CS050-10GM 5 MP mono GigE cameras with 25 mm lenses (MVL-HF2524M-10MP) and 4 × MV-LLDS-H bar lights | See the whole can body for visible dents |
| Processing (dent option) | HIKROBOT VC3000 vision controller (e.g. MV-VC3704H) running HIKROBOT VisionMaster with the deep-learning licence | Classifies dented and good cans from the four views |
| Trigger | HIKROBOT MV-PE photoelectric sensor | Fires the cameras when the can is in position |
| Reject | Not HIKROBOT: stainless air knife on a fast-switching valve, reject guide and bin; the code-only station signals your existing reject | Removes failed cans without stopping the line |
| Confirmation, tunnel and panel | Local supply: confirmation sensor, light tunnel and stainless mounts, panel with UPS and operator screen | Proves the can left; shields the side views from plant light |
Software and vision tools
| Tool or software | What it does here |
|---|---|
| OCR and character comparison on the vision sensor | Reads the code on the can end and compares it with the expected date and batch. |
| HIKROBOT VisionMaster OCV | Verifies each printed character against the expected string, so a broken character fails even when the code can still be guessed. |
| VisionMaster deep-learning defect detection | HIKROBOT shows deep learning finding defects in inkjet code characters on food packaging; trained on your good and bad codes. |
| VisionMaster deep-learning classification | Sorts side-view images into dented and good cans; trained on your own dented cans. |
| Expected values from the PLC | The PLC sends the date and batch for the run, so the check follows every code change. |
| Result and image log | Reject reason per can and saved images of failed codes and dents. |
Installation and integration
01
Mounting and enclosure
Vision sensor on a bracket over the conveyor straight after the coder; side cameras in a closed light tunnel on stainless mounts, clear of the can guides.
02
Lighting and shrouding
A dome light or ring light and a shroud over the can end, because a bright metal end mirrors anything above it; bar lights inside the tunnel for the side views.
03
Trigger and tracking
One trigger per can; the PLC or the vision system tracks each can from the cameras to the air jet so the right can is blown off.
04
PLC link
Pass, fail and reason over digital I/O or a fieldbus the chosen device lists; the PLC sends the expected date and batch at every code change.
05
Reject and confirmation
Air jet with a confirmation sensor and a full-bin signal; if the vision system is not ready, cans are rejected or the line is held, not passed unchecked.
06
Recipes and operator screen
One recipe per can size and code layout, chosen by the PLC at changeover; a screen shows the last rejected codes and dents.
// Indicative cost in South Africa
What this station costs, excluding VAT
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.
- Low
- R 100 000
- code only
- High
- R 1 180 000
- code plus 360° dent check and own reject
- Basis
- Typical application of this kind, indicative
- South African rand, excluding VAT
- Prices as at 2 October 2026
- Travel (Gauteng): outside Gauteng quoted separately
// Included and excluded
The low figure reads and checks the code on the can end with one vision sensor and signals your existing reject; it does not look for dents. The high figure adds four side cameras with deep-learning classification for visible dents, and an air-jet reject with confirmation. If the sample test shows your dents need 3D height data, a laser profiler replaces the side cameras; 3D work sits outside the standard ranges on the cost guide and is priced separately. 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 50 100 | R 410 900 | Low: one 5 MP vision sensor with OCR and a ring light. High: the same code station with a dome light, four 5 MP side cameras with lenses and bar lights, a VC3000 controller, stainless air knife and fast valve |
| Software licence | R 0 | R 120 000 | Vision sensor software is free; high case with the HIKROBOT VisionMaster deep-learning licence for dents |
| Local materials and reject | R 10 000 | R 140 000 | Low: bracket, shroud, cabling, signal to the existing reject. High: light tunnel and stainless mounts, reject guide and bin with confirmation, panel with UPS and screen |
| Engineering (recipes, PLC link) | R 21 600 | R 260 000 | 24 h at R 900 to 200 h at R 1 300 |
| Installation, FAT, SAT, training, documents | R 14 400 | R 124 800 | 16 h to 96 h |
| Project management | R 5 400 | R 52 000 | 6 h to 40 h |
| Travel (Gauteng) | R 1 500 | R 8 000 | Outside Gauteng quoted separately |
| Total (rounded) | R 100 000 | R 1 180 000 | Low: code only. High: code plus 360° dent check and own reject |
// Related technology
Which technologies this scenario uses, and why
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 core job is reading the code on every can end and checking it against the run’s date and batch. | OCR/OCV inspection |
| Vision sensor (HIKROBOT SC1000–SC3000) | Core | A code-only station needs no PC: a vision sensor with OCR over the conveyor does it. | Choosing a smart camera |
| Lighting technique | Supporting | A shiny, curved can end reflects everything above it; a dome or ring light and a shroud make the code readable. | Machine vision lighting |
| Deep learning (classification, segmentation, DL OCR) | Supporting | Character-defect detection for smudged inkjet codes, and classification of dented cans, both trained on your samples. | Rule-based vs deep learning |
| Area scan cameras with a vision controller | Supporting | Four side cameras on one controller cover the whole can body for dents. | Camera and lens selection |
| 3D laser profiling | Supporting | If shallow dents do not show in 2D images, a 3D laser profiler measures the can surface instead. | Dimensional inspection |
| PLC link and reject with confirmation | Supporting | Cans travel close together; the air jet must hit the right can and the reject must be confirmed. | PLC and reject integration |
// Acceptance
How the station is proven before handover
Every station is accepted on your products, against numbers agreed in writing before the build. How we deliver →
01
Sample test
Good cans, cans with missing, smudged and wrong codes, and real dented cans from your line, imaged with the proposed cameras and lights; the result says which codes and which dents are reliably found and which are not.
02
FAT with challenge cans
Before delivery, marked challenge cans (no code, broken code, wrong date, dented) must each be rejected, and good cans of each size must pass.
03
SAT on the line
On your line, at production speed, the challenge cans are run again, and false rejects and escapes are counted over an agreed run of production.
// FAQ
Questions about can inspection
What does a can inspection system check?
A can inspection system, sometimes called a can inspection machine, checks two things from outside the can. It reads the date and batch code on the can end and compares it with what the line should print, and, on the fuller station, it looks for visible dents in the can body from four side views. Cans that fail are blown off the line and the reject is confirmed.
Can a camera find dented cans?
Visible dents in the can body, yes, when the dent changes how the body reflects light; deep-learning classification is trained on your own dented and good cans. HIKROBOT publishes no dented-can case, so we prove it on your samples first. Shallow dents may need a 3D laser profiler, which is priced separately.
Is this can seam inspection equipment?
No. Can seam inspection equipment measures the double seam, usually on cans taken off the line and cut open. A camera looking at the outside of the can does not see seam faults, pinholes or leaks. This station checks the code and visible dents; seam checks and leak testing stay in your QA routine.
How much does can inspection cost in South Africa?
An indicative installed budget is R 100 000 – R 1 180 000 excluding VAT, at 2 October 2026 prices. The low figure is one HIKROBOT vision sensor reading the can-end code and signalling your existing reject. The high figure adds four side cameras with deep learning for dents and an air-jet 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.
Will it read codes on both beverage cans and food cans?
Yes, with a recipe for each. A beverage can end and a food can end differ in size, shine and where the code is printed, and codes may be inkjet-printed or stamped. Each can size and code layout is taught from your samples, and the light is chosen for the end you run.
Does it check that the code is the right date, or only that a code is there?
It checks the content. The vision sensor reads the characters and compares them with the date and batch the PLC sends for the run. A can with yesterday’s date, or another line’s batch, fails even when the print itself is perfect.
Sources
- Nampak diversified metals: food cans and ends, aerosols and metal containers (company page, checked 2026-10-02)
- Nampak beverage cans and ends (plants in Johannesburg and Pretoria) (company page, checked 2026-10-02)
- Regulations: grading, packing and marking of canned vegetables (GN R1532 of 2003) (gazette (2003 text; later amendments not checked), checked 2026-10-02)
- Regulations governing general hygiene requirements for food premises (R638, 2018) (gazette, checked 2026-10-02)
- NRCS presentation: typical canned-food non-conformances (seam defects, leaking cans, pinholes) (regulator presentation, checked 2026-10-02)
- HIKROBOT Machine Vision Smart Product Catalog 2025Q4: VisionMaster OCV and defect tools (vendor catalogue, checked 2026-10-02)
- HIKROBOT Smart Product Catalog 2025Q4: spray-code defect detection on food packaging (case) (vendor catalogue, checked 2026-10-02)
Related:
More line solutions in food and beverage:
Returnable crates →Multipacks →Juice box straws →Dairy cups →Bread bag closures →Meat tray labels →Wine labels →Sachets and pouches →Staple food bags →Egg cartons and trays →Bottled water (PET) →Food tray seal inspection →Food packaging version verification →Bakery product vision inspection →Returnable glass bottle inspection →
// Next step
Send us your worst codes and a few dented cans
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.