Skip to content

DOKTEK VISION · VISION-F

Engineered per application

Food Vision Sorting & Inspection

DOKTEK's vision platform inspects each product individually — colour, size, shape, surface condition, orientation — and separates it by class with robotic or pneumatic handling. It is a demonstrated capability engineered for each product and line, not a catalogue sorter.

Demonstrated in DOKTEK's own robotic sorting footage on apricots and dried figs; every new product is engineered and validated as its own application.

DOKTEK's robotic sorting system selecting apricots and dried figs — the same machine-vision and robotics platform, applied to fruit sorting.

Machine summary

Status
Engineered per application — demonstrated on apricot and dried fig
Detection
Colour, size, shape, surface defects, orientation, classification
Separation
Robotic pick-and-place or pneumatic rejection, per application
Evidence
DOKTEK robotic sorting demonstration video
Accuracy
Validated on the customer's product; no generic figure is published

What a Vision Sorting System Does

Product passes a camera station under controlled illumination. Detection models establish each piece's position and orientation, measure its size and shape, and classify its colour and surface condition. The decision for that piece — accept, grade, reject — is then executed by a separation stage synchronised to the conveyor.

The value is per-piece judgement at line speed: the same criteria applied to every product, on every shift, with the criteria documented rather than carried in an inspector's head.

  • Object detection and keypoint detection
  • Position and orientation
  • Defect detection and surface condition
  • Colour and size classification
  • Robotic or pneumatic separation

Demonstrated Applications

DOKTEK's published robotic sorting footage shows the platform selecting apricots and dried figs — real product, DOKTEK's own machine, the same vision and robotics used on the eggplant machine applied to fruit. The apricot and dried fig pages describe each application in more detail.

  • Apricot sorting and quality classification
  • Dried fig inspection and sorting, including UV screening support
  • Vegetables and other fruit — evaluated with samples

Engineered per Application

Sorting is never generic: the illumination, the camera placement, the classes, the reject criteria and the separation method all follow from the product and from what the downstream process needs. DOKTEK engineers each system around those facts, which is why the platform carries the status engineered per application rather than production machine.

Performance is validated on the customer's product with the customer's criteria. No generic accuracy or throughput figure is published because none would transfer honestly between products.

What to Send for a Vision Review

A vision application can be assessed quickly from samples and a description of the classes the plant needs.

  • Product samples covering the full quality range, including rejects
  • The classes or grades required
  • Current inspection method and operators
  • Target throughput and line layout
  • Downstream process

What is confirmed today

Confirmed: vision-guided detection and robotic handling run on DOKTEK's production eggplant machine, and robotic sorting of apricots and dried figs is shown in DOKTEK's own demonstration video.

Not published: accuracy rates, false-reject rates, throughput and prices. Each is validated per application on the customer's product.

Frequently asked questions

Questions before a machine evaluation.

01

Is the vision sorting system a finished machine?

It is a demonstrated platform capability engineered per application. Cameras, illumination, classes and separation are configured for each product and line, then validated on the customer's product.

02

Which products has it been demonstrated on?

Apricots and dried figs, in DOKTEK's own robotic sorting footage, alongside the vision system running on the production eggplant machine. Other products are evaluated with samples.

03

Does it use AI?

DOKTEK uses deep-learning detection models for product and feature detection. The system is not self-learning or autonomous: classes and criteria are engineered, validated and documented per application.

04

Can it replace manual quality inspection?

For the criteria a camera can see — colour, size, shape, surface condition — the system applies them consistently at line speed. Which criteria that covers for a given product is established during the application review.

05

How does an optical sorting machine work?

Product passes a camera under controlled light; software detects each piece, measures it and classifies its colour and surface condition against defined classes; a separation stage synchronised to the conveyor then accepts, grades or rejects that piece. DOKTEK engineers each of those steps — illumination, cameras, classes, separation — for the specific product.

06

How much does a fruit sorting machine cost?

No generic price exists, because a vision sorter is engineered per application: the number of camera stations, the illumination, the separation method and the throughput all follow from the product and the classes. A quotation follows an application review on your samples.

Next step

Send product details for an engineering review.

Share product photos, a short production video, your current method and target output. DOKTEK FoodTech will confirm what can be evaluated for your line.