Bad recognition with good input image

[Austin Seymour]

Working on a program to pull data from a game. Have it polarized and cleaned but yet I still can't get a good read. It is low resolution but in a few tests scaling it up in photoshop did nothing.

I've ran it though several online OCRs and they always seem to be able to get it perfectly.

This is what I get more or less.

Reserved 267
Mana 702{969
Rage 50{50
Life 981/2,181
Shield m
Ward 267

it seems fully incapable of seeing that 221 as anything I try ends up with random letters or nothing at all.  The brackets instead of slashes I could deal with but occasionally it will turn them into numbers or not see them at all which causes issues. The list of issues goes on and and, leading me to believe I'm just doing something wrong.

I know its possible to take this image and get a perfect read because websites that claim to use Tesseract can do it.

[Zdenko Podobny]

please read and follow the docs: 

https://github.com/tesseract-ocr/tessdoc/blob/main/ImproveQuality.md

You can easily get output like this:

Reserved 267
Mana 702/969

Rage 50/50
Life 981/2,181
Shield 221

Ward 267

Zdenko

št 24. 11. 2022 o 7:00 Austin Seymour <fridg...@gmail.com> napísal(a):

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-oc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/43b2be8a-af6e-4578-83e6-0a57e6906a6cn%40googlegroups.com.

[jannes hoekman]

We have a not free postprocessing software with 38 filters for a optimal OCR Its called BIQE = Batch Image Quality Enhancer which also works with Tesseract 4 OCR 

Op donderdag 24 november 2022 om 20:57:17 UTC+1 schreef zdenop: