OCR KO since the 21st of July 2018 for a given type of image


#1

I used the OCR API every day without any problem until July 21, 2018 for this type of image:
ligne

Since that day, it does not work anymore (through the Free OCR API and through the Online OCR), the text is not recognized. (In this example, “85073aa294aaa16” is expected.)

What changed ? What should I change to make it work?


#2

The free OCR API endpoint has a problem with very small images since the update on July 21, we are working on a fix for this already. (The PRO API endpoints are not affected, this is one of the reasons why we defer upgrades to the PRO endpoints.)

But there is a good workaround: Please add scale=true to your api call, then all should work fine again. You can test this on the online ocr page, which uses this parameter internally.


#3

Thank you for your quick reply but I don’t understand it: I wrote in my first message that it does not work anymore through the Online OCR. So how can I test your workaround on the Online OCR page ?
(You can test it yourself with the image given in my first message: it does not work.)

Nethertheless, I made a test with the API adding the “scale=true” parameter, it does not work either.


#4

Hmm… I see. We will test it more on Monday.


#5

This issue has been fixed with today’s Free OCR API upgrade :smiley: