search

TheJoeFin / Text-Grab

Use OCR in Windows quickly and easily with Text Grab. With optional background process and notifications.
https://www.microsoft.com/en-us/p/text-grab/9mznkqj7sl0b?cid=TextGrabGitHub
MIT License
3.33k stars 223 forks source link

Bad text recognition - "its" polish/english languages #480

Closed mir77777 closed 4 months ago

mir77777 commented 4 months ago

Describe the bug

"its" return as "iłs" text_grab_wrong_ocr

Where is the bug

To Reproduce in windows I have 2 languages installed, English and Polish I tried to select Polish or English, but it did not help Grab Frame on the image attached

Expected behavior Should return “its”. A good option would be to use only 1 language

Where did you get Text Grab?

Desktop (please complete the following information):

TheJoeFin commented 4 months ago

This is a bug with the underlying OCR models being used. I am eager to use updated models when they become available to 3rd party developers, but currently the best way to improve would be to use the Text Grab settings to install Tesseract and use those models.