Help with unicharambigs

[rkvsraman]

Hello,


I am testing the tamil data set and i am getting following issue

பெயர் is getting recognised as டுபயர்

and யான  as யஈன

I created the following unicharambigs file but it isnt helping.

v2
டுப    பெ    1
ஈ    ா    1

I tried both 1 and 0 as type indicator but no use

I have attached source text, image and OCred text

I tested it with original tam.traineddata too with same result.



Any suggestions?

[ShreeDevi Kumar]

It has been a while since I looked at Tamil.

Please see https://sourceforge.net/projects/tesseracthindi/files/Tamil%20Training%20Files/

You may want to try v1 style of unicharambigs file to see if that works. 

Not sure if v2 is supported ...

https://github.com/tesseract-ocr/tesseract/issues/165

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

--
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-ocr+unsubscribe@googlegroups.com.
To post to this group, send email to tesser...@googlegroups.com.
Visit this group at https://groups.google.com/group/tesseract-ocr.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/0f16d94f-8cc7-4c50-9988-3b1eaacb333e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Quan Nguyen]

The unicharambigs file has 5 columns.

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract#the-unicharambigs-file

[ShreeDevi Kumar]

Quan,

The page also says ...

Version 3.03 and on supports a new, simpler format for the unicharambigs file:

v2
'' " 1
m rn 0
iii m 0

But it may not be supported as per post by zdenko on the issue.

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

--
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-ocr+unsubscribe@googlegroups.com.
To post to this group, send email to tesser...@googlegroups.com.
Visit this group at https://groups.google.com/group/tesseract-ocr.

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/a20eea05-48a5-456f-9ac7-edcb79e7822f%40googlegroups.com.

[rkvs...@gmail.com]

I tried v1 too. Wasn't much useful 😞

 

Sent from my Windows 10 phone

--
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 post to this group, send email to tesser...@googlegroups.com.
Visit this group at https://groups.google.com/group/tesseract-ocr.

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/0f16d94f-8cc7-4c50-9988-3b1eaacb333e%40googlegroups.com.

[nkantan r]

i am also interested in using tesseract ocr for e-pub conversion of some of scanned books that i have;  i used python for auto correction and also a dictionary is required for updating auto correction of mistakes like யஈன

regards
rnk