Ignore Layout

  • Last Post 13 July 2013
vivek posted this 29 June 2013

Hello OCR_SDK: We have an application where we want to get the Text Left to Right and Top to Bottom - we do NOT want to retain the layout / borders / spacing. I see that there are some other's who have asked a similar question. Is there a Parameter/Flag that ignores the layout ?
thanks, vivek.

Order By: Standard | Newest | Votes
Anastasia Galimova posted this 02 July 2013

Hello vivek,

Thank you for your question. Unfortunately, there is no such flag in Cloud OCR SDK. In FineReader Engine it is possible to put a text block on the entire image, in this case the text is recognized from left to right and from top to bottom.

We are consulting with our analyst, if this feature could be implemented in Cloud OCR SDK, and will reply you next week.

At this moment we can recommend to export the recognized words with its coordinates to an XML-file and put the words in the necessary order on your side.

  • Liked by
  • vivek
Anastasia Galimova posted this 13 July 2013

Unfortunately, this functionality (exporting the recognized text in the Left to Right and Top to Bottom order) should not be implemented soon. Now you can either get the text in the order that was detected automatically, or perform export to XML and process words with its coordinates on your side.

vivek posted this 23 July 2013

Anastasia: Thanks for looking into this. I see a couple more Customers on this forum that have similar problem like us: - Processing receipts with ProcessImage - Hopping in a Table of Numbers I think that if you ignored horizontal/vertical lines, tabs, spaces - and just processed line by line (left to right) it would solve a majore portion of our problems. Is there any way you can push this up in the Product plan ? thanks in advance, vivek.

SDK_support posted this 23 July 2013

Dear Vivek,

Thank you for the feedback! We would suggest you to vote this feature request up, if it is important for you. Meanwhile you can use, for example, processTextField method, where the whole image will be presented as one text block. Probably this approach will helps you to resolve the issue.

Best regards, Anastasiya.

  • Liked by
  • vivek