Station

Label verification station.

One smart camera with integrated lighting at the labeler exit checks every can right after the label goes on. An inductive proximity sensor triggers the camera off the can, so a loose label flap can’t false-trigger, at line speeds of 300 to 1,200 cans per minute.

Tell us about the system Read the article

Full write-up: Label verification at the labeler exit

The setup

How is the station set up?

One smart camera, one trigger per can, and a reject that removes the exact can that failed.

Camera

One smart camera at the labeler exit

A smart camera with integrated lighting is mounted right where the label has just gone on. Camera, light and processing share one housing, and a labeling problem is caught before the can reaches the case packer.

Trigger

A proximity sensor on the can

An inductive proximity sensor triggers the camera off the metal of the can, not the paper. A loose label flap can’t fire an extra trigger, so each can produces exactly one.

Speed

300 to 1,200 cans per minute

The station runs at line speeds from 300 to 1,200 cans per minute. 1,200 cans per minute is the ceiling for code-reading tasks like this one.

Reject

Encoder-tracked air reject

A can that fails goes to an encoder-tracked pneumatic reject, driven by a PLC output, which removes the exact can that failed.

On every can

What does it check on every label?

Presence

Did the label land flat?

The station confirms a label is there and lying where it should. A flap from a label that lost its glue fails.

Code

The printed code, character by character

OCR reads the code printed on the label one character at a time. Two products can share a label design and differ by a single letter or digit.

Match

Checked against the live order

The code is compared with the product the line is running, which the plant’s MES sends over EtherNet/IP or TCP/IP at changeover. Nobody types a code in, so last order’s code can’t stay in place by mistake.

Reports

What does quality get from it?

The station runs on the same JM Automated Solutions HMI as our other systems: live image, pass/fail, false-reject counts and the current recipe.

Reports

Daily and monthly reject reports

Rejects broken down by reason, so quality can see whether loose labels or wrong codes are the bigger problem.

Studies

Saved images and timestamped logs

Saved images and results let the plant study a problem after the fact, such as loose labels after a glue change, with timestamped logs for traceability.

Why it matters

The last cheap place to catch it

A wrong label is mislabeled product. A joint GMA, FMI and GS1 US survey (2010) put the average direct cost of a food recall at about $10 million.

Read the article

How does this station work, in full?

Label verification at the labeler exit The full write-up: why the labeler exit is the place to check, what the camera checks on every label, the trigger, live changeovers from the MES, and the reports. Read the article →

Next step

Start with your line.

Tell us what you make and what is getting through. Start with a feasibility study.

Every station is tuned and reported on two numbers kept apart: false rejects, good packs thrown out, and escapes, bad packs let through.

Tell us about the system

Guides: Fixing false rejects · Feasibility check · Rules or deep learning · Lighting mistakes