Unifons Vision · pallet scanner
Still scanning every pallet by hand?
A wrong pallet label means returns, angry customers and searching. The scan station reads every label on two sides, checks it against the order through your ERP and lets nothing wrong slip through unnoticed.
product visualisation: the scan station, one scan head on each side of the pallet · click the clip to pause
What happens to your pallet
How it works
The five control gates
Our principle: a wrongly read barcode that gets through is worse than a missed one. So every read must pass five independent checks.
Within one image
If reads within the same image contradict each other, the read is uncertain.
Between images
The geometry of the rig is measured: a label must reappear between shots where it should be.
Two independent readers
Two different reading engines must agree, or a second shot must confirm.
Second read on doubt
If one engine misses something, a targeted re-read happens before any verdict.
Checked against the order list
Every read label is checked against the order through your ERP, within the same 10 seconds. Anything that does not match comes back per label on the screen.
A human where it matters
Anything uncertain goes into a queue: the operator sees the image, corrects or confirms, and the line keeps moving. A held read is not an error: it's exactly what the system is supposed to do. Both numbers, automatic and held, we discuss openly.
For the deep reader
Vision is a standalone machine: three industrial cameras per side on a moving scan head, polarised lighting, its own computing. Recognition runs entirely on the station's own computer: no cloud, and your production data never leaves the premises. Only the link to your ERP uses your network.
A read that is not certain is not passed on: it goes to a human with its photo, and the human decides.
neural network on your own labels · five gates · audit trail · local, no cloud · measurement method →
Five fair questions
“Our labels are too bad: wrinkles, foil, glare.”
Standard readers are trained on clean labels. This network is trained on your own labels, with your printers and your wrinkles. Polarised light takes the glare off the stretch wrap.
proof: the label check: you see on your own photos what is read and what is not.
“A camera misreads now and then too.”
One camera with one reader does. Here two independent readers must agree or a second shot must confirm, and the label must be on the order. Anything uncertain goes to a human, photo included.
proof: every installation is tested on zero wrongly passed labels. how we measure →
“That will slow the line down.”
Don't compare it with zero seconds, compare it with today's hand scanner. The station does a full pallet on two sides in under 10 seconds, including the check against the order.
proof: time one pallet by hand and put the two side by side.
“That turns into an IT project with our ERP.”
The station needs two things: the order in and the result out. Recognition runs on the station's own computer, no cloud.
proof: the platform connects to Exact Online, SAP Business One and Microsoft Dynamics 365.
“And what if it doesn't work for us?”
That is why it starts small. First the label check on photos, then an assessment on your site. Delivery ends with an acceptance test on your own pallets: one wrongly passed label and the test is failed.
proof: the test is described on the page how we measure →
THE FIRST STEP
The free label check
You don't have to buy anything to know whether it works for you. Send photos of your own labels and you get a read report back.
you get a reply from a human, no autoreply
No more wrong labels out the door?
Start with the free label check: about 20 photos of your labels, a read report back.
Request the free label checkReinier Weerts
founder, hands-on technologist
you get a reply from a human: no forms, no ticket numbers