FIRECRUX X-Ray QC
AI that finds contaminants and packaging defects on X-ray images, right on the production line: stainless steel, ceramics, glass, mineral stone, calcified bone and plastics, but also open or weak seals and cracked glass. It tells you what the defect is, not only that it is there.
Private data cloud for Industry 4.0 manufacturers.
AI solutions and models based on your data.
Unified workflow for production automation engineers.
Horizontal and vertical systems and machines integration.The proof is in the numbers
contaminants detected
production line speed
false alarms
trays scanned since 2019
Three ways in, and you choose how far to go
All prices net. A PoC tells you, on your own products, what the model catches before you decide on anything else.
Proof of concept
Send us samples of your products, including some with the defects you want to catch. We scan them in an X-ray lab, train a model and show you what it found and what it missed.
Our AI on your X-ray machine
You already have an X-ray inspection machine that misses too much or stops the line too often. We connect our AI to it, integrate it with your line and train it on your products.
The full line
We deliver everything: the X-ray machine, the industrial computer, installation, calibration, AI training on your products and a lifetime license.
What it finds
Contaminants
Stainless steel from 0.5 mm, glass, ceramics, mineral stone, calcified bone and plastics denser than the product, each recognized by type and size.
Packaging defects
Open or weak seals on flowpack packs, tested on ice cream.
Glass damage
Cracked glass under the bottle cap, tested on bottles.
Completeness
The number of elements in a set and how a filling is placed, tested on biscuits. The same models count, measure and compare products with a pattern.
What classic X-ray inspection cannot do without AI
Knows what it sees
Recognizes the product on the belt and the type of contaminant on the fly, not only that something is there.
Recalibrates itself
When the product or the conditions change, it recalibrates in under a second, with no technician.
One machine, many products
The same X-ray device checks different products, instead of one machine per product.
Real time, any working area
Detection runs in real time across the whole scanning area, whatever its size.
X-Ray QC questions and answers
How does Firecrux X-Ray QC work?
Two neural networks look at every X-ray image: one trained on clean products, one on products with defects. When they agree, the product passes or goes to the bin. When they disagree, the case is flagged for a person.
Can it catch a contaminant it has never seen?
Yes. The model is trained on the perfect product too, so a scan that is not confirmed clean is stopped, even when the contaminant is outside the tested range.
Does it work with the X-ray machine we already have?
Yes. It works with the X-ray generators and line-scan detectors already on your line. At Milarex, it runs with Spellman monobloc generators and Detection Technology line-scan detectors.
What hardware does it need?
An industrial computer next to the line. The models run on GPU, CPU or VPU, and a central server manages every machine and collects the detection statistics.
Does a detection stop the line?
Not with continuous scanning: a contaminated product is moved off the belt while the line keeps running. The system also works with cyclic machines that scan a batch of trays at a time.
How do the models get better over time?
Operators mark false alarms and confirmed detections, and the models keep learning from them. Your production data trains models built for your products.
Where are our X-ray images stored?
Your images stay in a private data cloud. At Milarex, it runs locally, at the plant.
What products has it been tested on?
It inspects salmon on 14 production lines at Milarex. We have also tested it on other food and drink:
- hazelnut shells in chocolate, for Mondelez;
- open or weak seals on flowpack ice cream packs, for a global ice cream maker;
- cracked glass under bottle caps, for Carlsberg;
- how cream is placed in biscuits.
Firecrux helped us develop the innovative technology in the industry supporting our food production process. Not only did they understand the problem, but also proposed a solution using machine learning algorithms and successfully implemented it in our factory. We are extremely pleased with the cooperation.Jerzy Malek, CEO of Milarex
What it changed for our clients
Firecrux X-Ray QCMilarex
99.9%Firecrux X-Ray QC at Milarex: 99.9% of contaminants caught in salmon processing, up from 62%, false alarms under 1%, lines at 1.2 → 2 m/s.
Read the case →
Firecrux X-Ray QCMondelez
85%+A test for Mondelez: the first X-ray AI models found over 85% of hazelnut shell fragments in chocolate, with well above 95% expected from more data.
Read the case →
Firecrux X-Ray QCCarlsberg
X-ray AI for Carlsberg that finds bottle mouths damaged during capping, even with the cap on, tested on 32 X-ray images of bottles.
Read the case →
Firecrux X-Ray QCA global ice cream maker
A test of X-ray AI on flowpack ice cream packs for a global ice cream maker: samples scanned in an X-ray lab, then a trial on a factory line in Poland.
Read the case →
Firecrux X-Ray QCEuro Industry
14licensesEuro Industry, a maker of fish processing machinery, runs Firecrux X-Ray QC on the ES 450, ES 600 and ES 800 machines we programmed together: 14 licenses.
Read the case →Plan X-ray inspection for your lines
Planning to deploy Firecrux X-Ray QC, or to improve detection on your lines? Complete the form and our engineers will contact you.
Your start with Firecrux X-Ray QC
- PoC on your samplesWe scan samples of your product in an X-ray lab and test detection before anything is installed.
- Insights delivery
- Development & deployment
- Full auto mode (AIRE)