OCR number issue?

  • Last Post 12 September 2013
DHDHDH posted this 04 September 2013

Take a look at this image: http://imgur.com/u1VAsex

Very simple image, OCR is returning 77266, it should be 11266.

Are there settings i can change? Are other users having issue with numbers should we look into a different OCR? This seems like a pretty basic task, that is the exact image i am sending for conversion.

Order By: Standard | Newest | Votes
Anastasia Galimova posted this 04 September 2013

We recommend to set language=digits. In this case the issue is not reproduced.

We could reproduce the issue when the English language is used. It is already fixed in the latest version of our technologies and will be fixed in Cloud OCR SDK about December, 2013 (after the technologies update).

DHDHDH posted this 05 September 2013

Even with Language Digits set, i am still having an issue take a look at this:


The originally image size is 270x62, i get a result of 7726, i try scaling down to 100x22 and still get 7726. Are numbers that problematic?

Anastasia Galimova posted this 12 September 2013

Most probably the issue occurs because the space between "1" is too large ("1" and "6" takes the same amount of space, as in fixed-width fonts), and for the used font "1" should take less space than "6".

If it is suitabe for your task, you can specify the coorditates of each digit.

This issue is also not reproduced in the latest version of the technologies and should not be reproduced in Cloud OCR SDK since about April 2014.