OKDate, batch and time
Reading (OCR) and verification (OCV) of best-before date, batch and time against the production order. Alert when the print head starts to degrade.
Home / Machine vision
100% in-line inspection
Machine vision to inspect products and packaging
No defective product should reach the customer. A cold seal, an illegible date or a case with eleven bottles won't show up in the hourly sample. They are only detected by inspecting every unit. That is the role of machine vision: inspect, measure and decide in milliseconds, and keep the proof. Millions of packs have already passed through our cameras.
What we inspect
Integrity, coding and counting in a single step
Most of a packer's complaints don't come from the product; they come from the pack. These are the inspections we install most often. If your defect isn't on the list, we'll run a remote feasibility test free of charge, using photos, videos or samples of your packs.
Cap and closure
Cap presence, height, centring and skew; opening tab and tamper seal.
Seal
Seal width and continuity, product trapped in the seal, creases and folds.
SKU and label
Correct label for the running format, presence, position and skew. Essential for private label and format changeovers.
Barcodes and DataMatrix
Reading and quality grading of GS1-128 and GS1 DataMatrix, and matching against the product.
Fill level
Underfill, overfill and empty or misfed packs.
Counting and grouping
Units per tray, case or shrink-wrapped pack before wrapping: not a single pack missing.
Pallet SSCC
SSCC reading, matching against the plant system and blocking dispatch if it does not match.
By sector: food, beverages and pharma. If your current system rejects good packs, read the guide to false rejects in machine vision. And if you are considering a new system, the guide how to choose a machine vision system covers what is worth asking before requesting quotes.
How it decides
One recipe per format, one verdict per pack
Each format has its own recipe: which zones to look at, with which tool and at what threshold. The camera scores each tool; if any falls out of range, the pack is rejected and its image is stored with the exact reason.
- Thresholds visible to the operator; only the quality manager and maintenance can adjust them, with user login and audit log
- AND/OR logic between tools for complex cases
- Format changeover by loading another recipe, in seconds
- Supervised bypass mode for maintenance
- Instant alerts by email, Telegram and WhatsApp
How it's built
Five parts and one sequence
The camera sees, the trigger decides when, the recipe sets the verdict, the rejector acts and the log keeps the record. If the timing between trigger and rejector is off, the wrong pack gets rejected. That's why we tune it on site, with real product.
- Lighting matters more than the camera. The lighting angle is what makes a contaminated seal visible.
- The trigger has to be stable. Photocell at medium speeds; encoder when the product shifts or the line runs fast.
- Rejection is measured in milliseconds. The delay depends on the distance and the conveyor speed.
TIVOX HUBLiveLive map · one report per minute from every line
Live map · one report per minute from every line
- Producing
- Attention
- Stopped
Mock-up with demo data: customers, lines and figures are fictitious.
Camera monitoring
All your cameras, watched from one place
Every vision system reports to TIVOX Hub on its own: whether it is producing, whether a camera has disconnected, whether someone has switched on bypass. Green, all running; amber, something needs attention; red, stopped or silent. We detect it instantly and alert you before it affects the line.
- Customer, line and camera, at a glance and in real time.
- Instant alerts if a camera disconnects or goes into bypass.
- The connection always goes out from your plant: your equipment is not accessed from outside.
Hardware
HIKROBOT cameras
Compact smart cameras with on-camera processing and integrated lighting, designed for presence detection, measurement, counting and reading of characters and codes. We choose the optics and lighting for each inspection point.
- Manufacturer
- HIKROBOT
- Camera software
- SCMVS + TIVOX recipe
- Integration
- Digital I/O, Modbus TCP
- Data
- Sentinel PLUS: image and verdict
Standards
We help you comply
We trace everything that passes our cameras. Every reject is stored with its image, time, recipe and reason, and NOK images are kept for the product's entire shelf life: the evidence audits ask for, and the evidence that settles a complaint months later.
- IFS · BRCGS
- Seal and labelling as control points
- HACCP
- Continuous records instead of sampling
- Reg. (EU) 1169/2011
- Legible date and batch on every unit
- GS1-128 · DataMatrix
- Reading and print grade
- Traceability
- Image of every rejected pack
- IP65 · IP67
- Enclosures for wet areas
FAQ
Answers to the most common questions
Which Spanish company does machine vision for packaging?
TIVOX Solutions, an engineering company in Madrid specialised in machine vision for food, beverage and pharmaceutical packaging lines, with 25 years of experience in the sector and systems in production in dairy, juice and broth plants. We design, program, install and maintain the complete inspection (date and batch, cap, seal, label and codes, counting) across Spain and Portugal, with our own software, Sentinel PLUS, for reject and OEE data.
What can machine vision inspect on a packaging line?
Date, batch and time (OCR reading and OCV verification), caps and closures, seals, labels, barcodes and DataMatrix, fill level, counting in cases and multipacks, and the pallet SSCC. On 100% of packs, in line, with a photo of every reject.
What is smart vision on a packaging line?
Vision with smart cameras: compact cameras that process the image inside the camera itself and have integrated lighting. They check presence, measure, count and read characters and codes on every pack, and decide in milliseconds whether it passes or is rejected. We choose the optics and lighting for each inspection point.
Classic machine vision or deep learning?
Classic vision when the defect can be described with rules: measurements, presence, character reading. Deep learning when it cannot: seals, surfaces or natural and highly variable products, with neural networks trained on your own products. Deep learning in inspection →
What if my defect is not on the list?
We run a remote feasibility test at no cost: you send us photos, videos or samples of your packs (if a visit to your plant is needed, travel is charged separately). We tell you whether it can be detected in line, with which technique and with what reliability to expect; what cannot be guaranteed, we tell you up front.
Where is machine vision used in industry?
Anywhere in the plant where a person would otherwise have to look: checking labels and codes, date and batch, caps and closures, fill level or seals, counting packs in cases and bundles, or guiding a robot to pick a part. It is used most on packaging lines, which run too fast to be checked by eye.
What are some examples of machine vision in a factory?
Three of our real projects: reading the code on every bottle and checking its cap in a dairy plant; detecting incomplete juice bundles with a single deep learning recipe for all references; and using 8 cameras before wrap-around case closing to check that every container is the same type and has its cap and code.
What happens in an automated visual inspection?
A camera takes a picture of every product as it passes, the software compares it with what it should be (text, position, shape or colour) and decides within milliseconds whether it is good or bad. If it is bad, a rejector removes it from the line, and the image is stored so you know what happened and when.
Free remote feasibility test
Can your defect be detected?
Send us photos of a good pack and a bad one, or real samples. We'll tell you whether it can be detected in-line, with which technique and with what expected reliability. Anything that can't be guaranteed, we'll tell you up front; whatever we install, we guarantee.