Add frequently used words to Tesseract

[Nilesh Gundecha]

Hi Friends,

I am using Tesseract 3.02 (with Tess4j java wrapper) on Windows 7.

I have got some list of words which are frequently used. I want to add these words in the Tesseract vocabulary.

I have gone through the TrainingTesseract3 wiki, but was just wondering in order to achieve this, do I need to create the tiff files, box files, .tr files, etc.

Cant I just generate eng.freq-dawg and add it to Tesseract vocabulary.

Any help is highly appreciable.

Thanks for this wonderful product.

Regards,

Nilesh

[Nilesh Gundecha]

I got the solution for this. All I had to so is - 

1) Unpack the eng.traineddata using combinedata tool

2) Add frequent words as per given in the Wiki

3) Pack all the eng.* files

However, I would like to add vocabulary to user-words rather than the freq-dawg. So I tried that as per it is given in the Wiki page.

But even after adding user-words file, there is no change in the OCR result. Any idea on this?

Regards,

Nilesh

[Ray Smith]

User-words gained an extra level of indirection. You don't have to put it in the traineddata file, but you have to specify an INIT parameter giving the extension of the user-words dawg. The INIT parameter can only be set via a config file provided with the call to Init. See http://code.google.com/p/tesseract-ocr/wiki/ControlParams

On Fri, Sep 21, 2012 at 10:24 AM, Zdenko Podobný <zde...@gmail.com> wrote:

Dne čtvrtek, 13. září 2012 11:24:14 UTC+2 Nilesh Gundecha napsal(a):

You do not need to run full training - pay attention to "Putting it all together" in TrainingTesseract3 wiki. BTW: in 3.02 version there is tool that creates wordlist from dawg dictionary, so you can merge your list with original.

--

Zdenko