Issues with OCR recogniztion

[Daniele Consoli]

Greetings, 
I'm new to Tesseract, but it seems like I'm having some fundemental issues and couldn't find any solutions by looking at guides online, so I figured I'd ask here.

I am trying to scan a few pages of a pdf, but the results are typically very inaccurate (see pictures below). 

I installed this version of Tesseract (https://github.com/UB-Mannheim/tesseract/wiki) for Windows.

The command I used to get those results is "tesseract --oem 3 5.png 5"

Any help would be greatly appreciated!

[Daniele Consoli]

Update, I have fixed this myself by using an app to turn all pictures black and white.
I think the issue might have been with the binarisation of the images that tesseract was doing.

I'm getting much better results now!

[Zdenko Podobny]

I would suggest reading the doc before using any tool. 

In this case: https://tesseract-ocr.github.io/tessdoc/ImproveQuality.html

Zdenko

so 10. 9. 2022 o 9:01 Daniele Consoli <consoli...@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/7d87a070-015e-49fc-9961-de3b0fa3b0f3n%40googlegroups.com.