How Cloud SDK OCR works with vertical and horizontal text?

  • Last Post 29 March 2012
rblasco posted this 19 March 2012

When I submit a document with horizontally and vertically oriented text (very usual in invoices), the OCR results seem to be incorrect (it only returns the vertically oriented text), is that right or am I doing something wrong?


Order By: Standard | Newest | Votes
Nikolay_Kh posted this 19 March 2012

Hi there! Please provide the document that is being processed incorrectly to so we could analize your case. Thanks!

Nikolay_Kh posted this 29 March 2012

Hi again, we've made up a two new options, imageSource and correctOrientation, the first one allows to turn off some image preprocessing options, the second one restricts auto-detect image orientation. We'll update documentation section shortly, but you can already use &imageSource=scanner&correctOrientation=false in your processImage, processDocument or processBusinessCard calls. Please let us know if you've worked it out!

rblasco posted this 29 March 2012

Hi Nikolay, I made a fast test yesterday, still have the same problem with the sample doc I send to you a few days ago but with another one containing vertically and horizontally oriented text the results where perfect. Maybe I should try with different parameter values or make some kind of preprocess before sending the ocr request to Abbyy. I will test it more exhaustively in the next days and let you know the results, thanks for your effort