Non-English image OCR doesn't work

Beta Beta Bug Reports
1

Testing version: 2.0 (9969)

What were you doing: OCR of images with text in different languages.

What feature did you use: Paste images → right-click → Copy image transcript.

What happened: OCR works correctly only with English text (tested Cyrillic and Hangul). Also app crash happens pretty often when you do that.

What did you expect to happen: You should be able to OCR such images.

To reproduce, create a note with the following 3 images:

english
cyrillic
hangul

OCR result:

this is english text (correct)
9TO pyCCKMi TeKCT (not correct)
(not correct, just no text)

Some additional things I noticed:

  • It looks like the app tries to use English letters which look similar to the letters on the image
  • Hard crash of the app is pretty often when you use the transcript feature
  • Search for notes with the text on non-English images also doesn’t work (seems like same issue)
  • Search for non-English text of PDF works (seems like different implementation is used for that)
  • I don’t have issues with non-English text recognition in native Apple apps such as Photos
2

Thanks for reporting this. We currently don’t support Korean and Cyrillic but we will with the next update.
We’ll look into the crash as soon as possible.

3

I can confirm: it works perfectly with 2.0 (10084).
Thank you very much!

(But to make search feature work, I had to re-paste images again, seems like indexation is done when you paste an image into note.)

1 Like
4

Hello,

Does it work for French? (I attached an example)

image
image1170×2532 148 KB

5

You possibly have to re-do the migration from Bear 1 to Bear 2 if you want notes search to work with attachments.

6

Yes it does. Are you experiencing otherwise with the image you posted?