Label Inspection System for Packaging Lines
Inline label inspection that checks presence, position, the right SKU label and legible batch codes on every pack, with pass/fail sent to your PLC.
A label inspection system checks every pack on the line for the right label, in the right place, with a readable batch code, and rejects the ones that are wrong before they reach a carton. DeepVision does this inline at line speed and sends each pass/fail decision to your PLC.
Figures from one deployment. Read the FMCG label inspection case study.
What a label inspection system checks
| Check | What goes wrong on the line |
|---|---|
| Label presence | A pack leaves the labeller with no label, or with a front label but no back label. |
| Position and alignment | The label is skewed, shifted or applied too high or too low. |
| Right label for the product | A label from another SKU or language variant is applied after a changeover. |
| Print quality | Batch codes, dates and prices are smudged, faint, partly missing or unreadable. |
| Application defects | Wrinkles, flags, bubbles and torn edges that make the pack look damaged on the shelf. |
| Seal and closure | On the same station, an open or badly formed seal can be checked alongside the label. |
Why manual and sample checks miss label defects
Spot checks only see the units that are pulled. A labeller that drifts out of position, a roll that runs low or a printer head that starts to fade can produce faulty packs between two samples, and those packs ship. Inline inspection looks at every unit, so an intermittent fault shows up as soon as it starts.
How DeepVision inspects labels
- Capture. A camera after the labeller images every pack as it passes.
- Compare with good product. DeepVision uses anomaly detection: it learns what a correctly labelled pack looks like from images of good product, so you don’t need a library of defect photos to start.
- Decide. Each pack gets a pass or fail decision on an edge computer next to the line.
- Reject. The decision goes to your PLC, which drives the existing reject mechanism.
- Record. Every decision is stored with its image, so a reject or a customer query can be traced back to the actual pack.
Where label inspection is used
- FMCG and personal care: bottles, tubes, jars and cartons. See AI inspection for FMCG.
- Pharmaceuticals: labels on bottles, vials and cartons, with an image-backed audit trail. See pharma inspection.
- Food and beverage: front and back labels, date codes and multi-SKU lines.
Frequently asked questions
Can it tell one SKU’s label from another?
Yes. Because DeepVision learns the correct label for the product being run, a label from a different SKU looks different from the good reference and is flagged.
Does it read batch codes and dates?
It checks that printed codes are present and legible. Talk to us if you also need the printed text read and matched against a value from your line.
How long does a deployment take?
On the FMCG line in our case study, the system was live on day 7 after installation. The time for your line depends on the number of stations, mounting and integration.
Do we need defect samples to train it?
No. DeepVision is trained on images of good product. Defect samples help with validation but are not needed to get started.
Labels escaping to customers? Read the FMCG case study, see DeepVision AI visual inspection, or book a demo with your own packs.