4 things to know about OCR - Understanding key automation technologies

October 19, 2020

Welcome to week four and the final installment of ‘Understanding key automation technologies’, this week our focus is Optical Character Recognition (OCR).

As with the other articles in this series, our aim is to provide bitesize introductions to the complicated technological methods used in the automation world.

All previous articles, covering Robotic Process Automation, Machine Learning and Cognitive Computing, can be found here.

Let’s dive into OCR.

Definition:

Optical Character Recognition is algorithms that identify text (printed and handwritten) in digital images (photographs, scanned documents).

OCR identifies the text and converts it

  • OCR technology recognizes characters (letters, numbers, symbols) in an image and converts the text within the image into ‘machine-readable’ text (meaning a computer would recognize the document as text, not an image), which can then be exported.
  • Some OCR software will simply export the text files, whereas others can make the text in the converted files editable, searchable and possibly even translate into further languages.

OCR Uses

  • OCR is often used to recognize text in scanned documents and convert this into text files, which can be very valuable for data entry scenarios.
  • One amazing use of OCR is the Google Translate app – it can identify and translate text both in photos and through the camera. It can detect languages, or you can select the languages ‘from’ and ‘to’ translation and it will provide the translation for you.

Advantages of OCR

Accuracy and Speed

  • With regards to scanning documentation, before OCR existed, it would have been a manual process to re-type and re-create the scanned document into another program (Word or similar) to make them editable. This is not only time consuming but likely to come with transcription errors.
  • Now, OCR can identify text in images and convert the document in seconds, (in theory); we’ll come back to this a little later.

Improve Productivity

  • In office environments with high volumes of paperwork (cross checking information, scanning documents etc.) OCR can provide immensely useful support and give hours back to the team to focus on pressing business areas that need their attention, such as problem solving, customer service and engagement.

Searchability

  • Once the text is lifted from the image and converted to be machine-readable, it can be searchable. This has been a revolution for digitizing newspaper archives and ancestry records, opening this information up to whole new generations.

Security

  • If documents can now be scanned and secured safely digitally, the physical paper copies may not be needed any longer. Therefore, the risks of physical copies getting into the wrong hands or being filed incorrectly are removed.
  • If companies historically outsourced documents to be manually re-created for digital use, with OCR this is no longer a requirement, so it ensures sensitive documents do not need to be shared unnecessarily with third parties.

 

OCR, when carried out effectively, comes with many benefits. However, not all OCR solutions are created equally, and some could still require a fair amount of human input. Many off-the-shelf solutions rely heavily on good quality scanned documents or the bots cannot read the documents effectively. High quality scanned documents are not always possible, especially with older records. If this happens it can lead to frustrations as the new technology is not working as hoped.

That’s why, at Roots Automation, we’re building our own OCR capabilities as we don’t believe customers should settle for forms to be read with only 70% accuracy. We are focusing our efforts on a subset of forms, which are very widely used in the US, and training our bots to read these forms - even with poor quality input and scanning- to over 99% accuracy, in seconds.

One example: Most OCR software would not count these boxes below as being checked but we have trained our bots to infer if the box is checked by looking at the data around the box. Here we show our tool identifying an ‘X’ and appropriately recognizing the box it belongs to, even though it was not in the box!

Could your business benefit from OCR, or any other automations we’ve discussed in this series?

Continue the conversation with us on Twitter and LinkedIn or reach out to us with any questions and to book your free demo here: info@rootsautomation.com.

Thanks for reading!

 

Subscribe to our blog