Deep Learning OCR

Deep Learning OCR from Zebra Aurora is changing the way we look at text recognition in industry

Imagine a system that can instantly read hard-to-read text on shiny metal, faded plastic or broken labels – without you having to teach it how. It sounds like science fiction, but with Deep Learning OCR from Zebra Aurora Vision™, it’s already a reality. This technology is revolutionizing the way industry works with traceability, quality assurance and automation – and you can implement it without writing a line of code.

Deep Learning OCS in Aurora Vision
barcode

What is Deep Learning OCR?

Deep Learning OCR is the next generation technology for automatically reading and interpreting text from images or video streams. Unlike traditionalOptical Character Recognition (OCR) tools that require manual font training and careful settings, Deep Learning OCR relies on neural networks trained on large datasets of different fonts, text styles and environments.

The result is a system that is quick to implement, easy to use and extremely robust – even in tough, unpredictable conditions.

What is Zebra Aurora Vision?

Zebra Aurora Vision™ is a machine vision platform that enables the development of visual inspection systems for industrial applications. The software is used in the manufacturing, logistics, automotive, medical and packaging industries, among others.

Aurora Vision includes a powerful graphical programming interface where you can build visual flows using modules and filters – all without writing traditional code.

The platform offers support for, among others:

  • Image analysis and quality control

  • Barcode reading and data matrix recognition

  • Pattern recognition and shape matching

  • Deep Learning OCR– a key feature for text recognition

Talk to Grafokett about Deep Learning OCR

Benefits of using Deep Learning OCR in Zebra Aurora

Zebra Aurora’s implementation of Deep Learning OCR provides users with a tool that is quick to implement, robust in harsh environments, flexible and customizable, and not least, very user-friendly.

  • Quick to implement

    No font training required – the pre-trained models are ready to use right away. You drag in the DL_ReadCharacters filter, plug in your image source and the OCR is up and running.

  • Robust in harsh environments

    The system works and reads characters even under the following conditions:

    • Damaged, skewed or low-contrast text
    • Reflective surfaces like metal and plastic
    • Transparent materials
    • Braille (dot matrix)
    • Unevenly lit backgrounds
  • Flexible and adjustable

    Although the basic settings work well right away, you can easily adjust:

    • inCharHeight: font size
    • inWidthScale: for wide or narrow characters
    • inCharRange: which symbols to include
    • Polarization: light or dark text
    • MinScore and Contrast Threshold: precision thresholds
  • User-friendly

    Thanks to the visual programming interface, no previous programming experience or deep knowledge of machine learning is required. This makes Zebra Aurora accessible to both machine vision engineers and production engineers.

Who benefits from Zebra Aurora
and Deep Learning OCR?

  • Manufacturing industry

    To read batch numbers, product codes, date marking or engraved text on components and packaging.

  • Automotive industry

    To interpret stamped or etched serial numbers directly on metal, plastic or rubber – even in dirty or poorly lit conditions.

  • Medical technology

    To automatically read labels and identification codes on vials, tubes, packages or implants.

  • Logistics and warehousing

    To read text directly on cartons, pallets and packaging where labels are sometimes missing or damaged.

  • OEMs and machine integrators

    For system builders who need fast and reliable OCR solutions to integrate into larger automation systems or robot cells.

Example of use

The Zebra video demonstration shows how you can set up an OCR system in Aurora Vision Studio in just a few minutes. By loading an image stream, adding DL_ReadCharacters, and optionally using MergeCharactersIntoLines to structure the text, the OCR is up and running.

It is demonstrated how the system effortlessly copes:

  • Dark text on light background (and vice versa)
  • Plastic material with embossed text
  • Reflective metal
  • Text that is crooked, damaged or low-resolution

You’ll also learn how to adjust parameters to optimize performance – all within an intuitive user interface.

Deep learning OCR in Zebra Aurora

Why Deep Learning OCR is the future

If you’re looking for a solution that combines deep learning OCR, industry-class reliability, and rapid development time, Zebra Aurora is one of the most complete tools on the market.

Those looking for solutions such as “Deep Learning OCR for industry”, “OCR that does not require font training” or “automatic text recognition on difficult surfaces such as metal and plastic” are probably looking for a future-proof and AI-driven image analysis platform. Zebra Aurora Vision meets exactly these needs – with an out-of-the-box OCR solution that is powerful, flexible and ready to integrate into modern automation environments.

Zebra Aurora Vision for e.g. OCR

Executive summary

Zebra Aurora Vision™ with Deep Learning OCR provides businesses with a powerful, user-friendly and scalable text recognition solution – ready to meet the automation, quality and traceability challenges of today and tomorrow.

You avoid complex code, don’t have to learn your own fonts, and get a solution that works in even the most challenging visual environments.

Want to learn more? Visit aurora-vision.zebra.com for technical documentation, guides and software downloads.

You are welcome to contact us for more information.

Frågor, funderingar eller idéer? Kontakta oss!