cloud.ocrsdk.com/processImage with fieldLevelRecognition

  • 234 Views
  • Last Post 17 May 2017
Dmitry Lapunov posted this 16 May 2017

Hello. I use C# samples. GUI example for windows,

Application call 

POST http://cloud.ocrsdk.com/processImage?language=english&exportFormat=xml&profile=documentConversion&textType=normal&description=123.png%20-%3E%20.xml

THIS WORKS.

I want recognition eaсh word with rectangle. I think i need use parameter profile as fieldLevelRecognition

http://ocrsdk.com/documentation/apireference/processImage/

 

I did POST http://cloud.ocrsdk.com/processImage?language=english&exportFormat=xml&profile=fieldLevelRecognition&textType=normal&description=Doc-1-1.png%20-%3E%20.xml 

but if i use fieldLevelRecognition, Responce from OCR CLOUD Service without result = Empty page.

<page width="2550" height="3300" resolution="310" originalCoords="1">

</page>

</document>

 

What do I did wrong?

tnx for answer.

 

Order By: Standard | Newest | Votes
Oksana Serdyuk posted this 17 May 2017

I would recommend to try field-level recognition, and recognize every needed word separately as a field. For this, you need to set up some certain recognition parameters that "tell" Cloud OCR SDK what kind of text is recognized and where it is located. This should be done for all blocks, because the automatic layout analysis is not available for field-level recognition. Please refer to the How to Recognize Text Fields article for an overview of possible applications. To extract the value of one text field on an image, you can use the processTextField method. And if you want to recognize a lot of small text fields on a page, we recommend to use the processFields method. You can see these methods’ implementation and test them in work using our C# GUI Sample. Select the "Field-level" tab. 

But please note that this approach is applicable, only if the recognized documents are well-structured, have a known layout and are not different from each and other. Otherwise it is better to use full-page recognition, export the result in the XML format and then apply some post-processing to extract necessary data.


 

Dmitry Lapunov posted this 17 May 2017

Hello, Oksana.

I tried to use

  • documentConversion -> big part text
  • textExtraction -> smaller part text but it isn word, it is some word and sentence  


How could i get from recognition system only word. I can get rows with sentence but i need word.

Thank you for help.

 

Oksana Serdyuk posted this 17 May 2017

Please do not use this profile, this is mistake that it is described in the documentation. This profile does not support layout analysis, so the result is always empty. We will update the documentation as soon as possible.

Sorry for possible inconvenience!

Close