Label inspection in the food industry: how to prevent recalls
A wrong label or an unreadable best-before date can bring back an entire production run. This is how machine vision checks every label, every code and every date on the line.
In the food industry, a wrong label is more than a cosmetic flaw. Missing allergen information or a wrong best-before date can lead to a recall, with costs for production, logistics and your reputation. The European rules on food information (Regulation (EU) No 1169/2011) leave little room for errors on the packaging.
Sampling helps, but anything that passes the line between two checks goes unseen. With machine vision, you check every product.
Where label errors come from
Most errors have a recognisable cause:
- Product changeovers: after a changeover the wrong label roll is in the machine, or the printer is still set to the previous product.
- The printer: a best-before date or batch code is faint, incomplete or shifted, for example because of a clogged print head or ink running out.
- The labelling machine: the label is crooked, creased or missing.
- Wrong combinations: lid, sleeve and label do not belong to the same product.
What a vision system checks
A vision system for label inspection and OCR looks at every product and checks, among other things:
- whether the label is present and straight in the right place;
- whether it is the right label for the product currently running;
- whether the best-before date and batch code are readable and correct (OCR);
- whether barcodes and 2D codes, such as EAN, QR and DataMatrix, are readable and correct;
- whether the print quality is good enough.
Connecting to your production
A check is only valuable when the system knows what the label should say. At a product changeover, the expected date or batch code comes from the line control or MES; the camera compares every product against it. If something deviates, a signal goes to the PLC, which rejects the product. Images of rejected products are stored, so you can show afterwards what was checked.
What matters in practice
- Shiny film and curved packaging cause reflections that make text unreadable. The right lighting is decisive here; see also why lighting matters more than the camera.
- Line speed determines how short the exposure must be and how much time the software has per product.
- Changing layouts per product or customer call for a system that can handle them. With AI, dynamic information can be recognised and validated too.
- Handling rejects: an ejector, a buffer and clear messages for the operator are part of the design.
From sampling to 100% inspection
Automated label inspection replaces sampling with a check of every product, records every deviation and acts before an error leaves the factory. Get in touch to discuss what that means for your line.