Low tesseract accuracy

[Dinesh Yakkanti]

Hello Everyone,

           I am trying to build custom tesseract-ocr model. I am getting high error rate even if i kept more iterations  ( 40k ). Below is the sample image i am using to train the tesseract-ocr model.  Can anyone suggest me how to improvise the accuracy and extract the text from the below image accurately. Please share me if anyone having any articles related to this and  research papers to implement.

Thanks & Regards,

Dinesh Yakkanti

[Shree Devi Kumar]

Suggest you pre-process images instead of training.

See https://tesseract-ocr.github.io/tessdoc/ImproveQuality.html

--
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/d030135a-7377-4c9b-95d9-2f82d0e9ecc2n%40googlegroups.com.

--

____________________________________________________________
भजन - कीर्तन - आरती @ http://bhajans.ramparivar.com