Automagic Orientation Detection with the new LSTM models?

[Breno Faria]

Hi everyone,

let me begin with a big fat compliment on the Tesseract project on the quality of the OCR since the new LSTM models have been adopted! You are consistently better than ABBYY now.

I have just made a few experiments with slightly rotated documents. There ABBYY is still better than Tesseract 4.

I was wondering if it wouldn't be very easy to just generate rotated training data by just duplicating and rotating the existing training documents. Wouldn't the model than automatically learn to handle this?

Has anyone tried this yet?

Cheers!

Breno

[Michael Smith]

I just do some preprocessing and auto rotate/clean up images before passing it on the Tesseract.

[Simon Eigeldinger]

Hi,

like me as a blind i wonder how i might use some of those tools?
because you can't see if the pic is good or bad.
actually we might need something that does that automatically.
any ideas on that?

Greetings and thanks,
Simon

---
Diese E-Mail wurde von Avast Antivirus-Software auf Viren geprüft.
https://www.avast.com/antivirus