Store
KI-Fremdstofferkennung auf dem Förderband mit Detection-Overlays und Klassifizierung
01

The problem

Manual visual inspections are error-prone, slow, and do not scale at high throughput. Contaminants such as metal, glass, plastic or paper in recycling streams or food lead to product recalls, equipment damage and fines.

<50msdetection latency on the belt
>99%detection rate on trained classes
02

The solution

KINEVA AI models analyse the conveyor belt flow via camera in real time. Detected contaminants trigger an immediate signal: ejection, belt stop or alarm. The system runs entirely on GUSTAV Edge Hardware, with no cloud connection and no network dependency.

Conveyor belt
KINEVA®
Ejection / Stop

Detected contaminant classes

The model is trained on your materials and production environment. Typical classes:

Metal

Screws, wires, metal chips and tool fragments in the material flow. Critical in food and recycling.

Glass and ceramics

Shards and splinters in bulk goods, food or recycling streams.

Plastic

Wrong plastic types in recycling sorting, or foreign objects in organic product flows.

Organic impurities

Mould, decay or out-of-spec material in food and agricultural product flows.

Benefits for your operation

No product recalls

Contaminants are detected before filling. No contaminated product leaves the facility.

Equipment protection

Metal objects and hard contaminants are detected before they damage machines. Unplanned downtime drops.

100% inline inspection

Every part, every unit, every second. No more sampling. Full inspection with no throughput loss.

No cloud required

Everything runs on the equipment. No data leaves the plant. No outage during network issues.

Detect contaminants reliably?

We analyse your production process and show how HORIZON Industrial safeguards your line.

Book a consultation now