OCR of a product code on an egg held by a hand.

AI text recognition with DENKnet

Recognise and process text on surfaces thanks to Deep OCR

Inspect labels, check delivery notes, read product numbers, and much more. OCR systems reliably recognise text on all surfaces and help to automate processes. Discover the unique features of DENKnet's Deep OCR software in terms of accuracy and performance here.

OCR of character strings on separator disks, displayed in the DENK Vision AI Hub.

Customised OCR systems for every use case

The DENK OCR software is based on artificial neural networks that are continuously evolving. The AI is trained using a combination of images from real applications, augmented data and synthetic data. This enables the system to handle special characters, difficult-to-read fonts, cluttered backgrounds and much more. Contact us to discuss your specific use case!

Close-up OCR of a label on a plastic bottle.

How does AI-based text recognition work?

OCR stands for Optical Character Recognition and describes the recognition of letters, characters, and numbers. In the industry, these are applied to various surfaces through printing, stamping, casting, or other marking methods.

For text recognition, an image of the respective surface is taken, where object recognition identifies the text. The OCR software then segments the text into lines, sentences, words, and characters. The AI evaluates the recognised characters based on learned features and outputs them accordingly.

The unique features of DENKnet Deep OCR software

DENKnets' AI endows text recognition with exceptionally high accuracy. The AI model is continuously updated with special cases and specific customer requirements. Augmented and synthetic data are automatically integrated during each training session. New knowledge is added and weighted without overwriting previously learned information. Furthermore, the OCR solution recognises patterns in character strings, such as item numbers, and incorporates this knowledge into further processing. Despite its high performance, the Deep OCR requires minimal resources, is easy to use, and operates quickly.

Find out more about deep OCR systems and their differences in our technical article.

A robotic finger is about to press a button labeled "OCR".

High performance with few resources

High accuracy

Continuous development

Knowledge retention

Easy to use

Customizable

pattern recognition

Discover the possibilities of OCR text recognition

With DENKnet, users can perform individual image analyses with leading AI technologies in just a few minutes. The combination of DENKnet image processing and IDS industrial cameras enables solutions of the highest performance and quality.

OCR of a ring-shaped embossing on a watch in close-up.
Ring-shaped embossing
OCR of a DOT number on a car tire in close-up.
DOT numbers on car tires
OCR of a dotted product code on a plastic bottle in close-up.
Dotted font on uneven surfaces
OCR of a serial number on a camera lens in close-up.
Text in small print on deeper surfaces
OCR of a best-before date on a can in close-up.
Shiny metallic and low-contrast backgrounds

How to read any text with DENK OCR

In our Vision Channel session, we demonstrate how you can achieve the best solution for your application using the DENK Vision AI Hub.