This support forum is no longer in use. Please use the Cloudinary Community at https://community.cloudinary.com/ to receive assistance from other members of the community and from Cloudinary's support team. For account-specific questions or if you believe that you've encountered a bug, please contact the Cloudinary team directly via the "submit a request" option on this support site.

Extracting cyrillic text using OCR plugin

Follow
Avatar
Dmitry Bezrukov

Hi,

I'm planning to use OCR text detection plugin in my project and I managed to extract text by running 'cloudinary.v2.api.update' with { ocr: 'adv_ocr' } parameter.
It extracts english text well:
http://res.cloudinary.com/godovod/image/upload/v1513136310/ocr/eng_bw.png
https://godovod.ru/api/files/ocr?file=ocr/eng_bw

But it doesn't extract russian, there is no 'ru' among detected lanuages:
http://res.cloudinary.com/godovod/image/upload/v1513138221/ocr/ru_bw.jpg
https://godovod.ru/api/files/ocr?file=ocr/ru_bw

Google Vision API itself extracts russion text perfectly:
https://gyazo.com/c652dba9a166e9f9058a60552b2d3a24

Are there any restrictions or maybe some OCR trained data or settings are missing?

1

Comments

2 comments

Sort by Date Votes
  • Avatar
    Dmitry Bezrukov

    I've made some tests - in order to extract russian text we should explicitly set 'languageHints' in Google Vision API requests when using TEXT_DETECTION feature. If we use more complex DOCUMENT_TEXT_DETECTION feature the API parses russian text without any hints. So in my project i started to use Vision API directly, passing url of Cloudinary-hosted images.

    1
    Comment actions Permalink
  • Avatar
    Ido

    Thanks for sharing Dmitry!

    0
    Comment actions Permalink

Post is closed for comments.