Machine vision that sorts at line speed
An industrial camera, a neural network and automatic real-time removal, running on an edge computer inside the machine, with no cloud dependency.
- Fiber purity
- up to98%
- Classification accuracy
- 96.4%
- Processing capacity
- up to3t/h
- Inference latency
- 7ms
See it, classify it, remove it, at conveyor speed
Flax is an irregular natural fiber: its colour, texture and density vary from bale to bale, and the contaminants hidden in it range from a strand of blue plastic string to a feather the same colour as the fiber. A sorting system has to cope with all of that without slowing the line.
SORTIX does it in five stages that run continuously while the belt moves. The fiber is imaged by an industrial camera under controlled lighting. A background-removal algorithm developed for natural fiber isolates the material from lighting noise. A convolutional neural network classifies what it sees across more than 15 contaminant types. The position of each contaminant is calculated from the conveyor speed and its time of arrival at the removal point. A timed automatic removal lifts it out of the stream.
Everything runs on an edge AI computer inside the machine: there is no cloud dependency and no internet connection required for sorting.
Automatic training on the factory floor: SORTIX keeps learning after installation. Images from your own line feed back into training on site, so the model adapts to your fiber, your lighting and any new contaminant, and gets more accurate the longer it runs.
Flax today. Any sortable material next.
In-line in flax processing today. The same inspection core, next for crops, food grains and recycling streams.
Available now
- Flax fiber processing and scutching
- Linen spinning preparation
- Export-grade fiber quality control
On the roadmap
- Agricultural crops
- Food grains
- Recycling streams
From raw fiber to clean fiber, in one pass
Five stages, all of them happening while the belt keeps moving.
Raw flax
Fiber enters on the conveyor exactly as it leaves the preceding stage; no change to your line is required upstream.
Imaging
An industrial camera captures the moving fiber under controlled lighting, frame by frame.
Analysis
Background removal isolates the fiber, then the neural network classifies what it sees.
Identification
The contaminant is located on the belt and its position is tracked from the conveyor speed.
Removal
The system removes it from the stream automatically, in real time, and clean fiber continues to the next stage.
Ultraviolet inspection for what visible light misses
Some contaminants are nearly invisible to a colour camera: clear film, white synthetic string, bleached fabric. Under ultraviolet illumination many synthetic and chemically treated materials fluoresce while natural flax does not. The SORTIX AI + UV machine adds that second imaging channel and fuses it with the AI vision channel before the removal decision: the same throughput, a wider net.
See it on your fiber
Send us a sample and your production quantity; we will run it and share the results.