shapeclustering, do or not do?

[Shane Wee]

I am using tesseract 3.0.2, I trained my data with shapeclustering included, the result is not as good comparing with the traineddata I got from excluding shapeclustering.

Shapeclutering seems to cause error recognition on similar shape character such as 1 and I, O and Q, 5 and S. 

I am quite sure I follow the training steps correctly.

My question is whether shapeclustering is really important? If I exclude it from my training, will I miss out anything important?

[Ray Smith]

The idea of shape clustering is that it should help to resolve exactly the errors that you observe! It doesn't work too well at the moment though for most languages. It currently should not be used except for the Indic languages, where it does seem to help.

Ray.

--
You received this message because you are subscribed to the Google Groups "tesseract-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-de...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Sriranga(79yrs)]

Tested for Kannada lang - shape clustering did not help for Kannada(one of the Indic) - vide attached files (1)with shapecluster- alphabet test.rtf and (2)without shapecluster -alpahabettest.rtf   no improvement at all.

[Shane Wee]

Thank you so much for the responses. In my case (training alphanumeric only), shapeclustering seems to make the results much worse than without shapeclustering. Thanks again for the confirmation.

[mamat...@gmail.com]

Sir
I have followed the procedure to train an indic language which is similar to bangla language using Tesseract-3.01.
But not sucess. Should I use Tesseract-3.02