mftraining produces "Missing font_properties"

2011-05-17 Thread Eyal
Hi, I tried to train some letters & when I ran the *mftraining *with the parameters*:* *mftraining -U unicharset -O lang.unicharset font1.tr *I recieved an error message: "Missing font_properties". I'm working on windows 7, visual studio 2010. When I used the already compiled mftraining.exe fo

Re: mftraining produces "Missing font_properties"

2011-05-17 Thread zdenko podobny
On Tue, May 17, 2011 at 9:08 AM, Eyal wrote: > Hi, > > I tried to train some letters & when I ran the *mftraining *with the > parameters*:* > *mftraining -U unicharset -O lang.unicharset font1.tr *I recieved an error > message: "Missing font_properties". > > I'm working on windows 7, visual studi

Re: mftraining produces "Missing font_properties"

2011-05-17 Thread Eyal
Quite a good guess, but I'm very disappointed to to say - I DID read the documentation... And I even run the following command: *mftraining -F font_properties -U unicharset font1.tr* And I got results which don't show any error... : *Reading font1.tr ...* * * *Writing Merged Microfeat ...Done!

About the jpn.traindata

2011-05-17 Thread Mostafa
Hi, I am interested to get all the tif files that used for creating the jpn.traindata. I just want to see how many characters are supported in that file. Because I have some other Japanese characters that can't be recognized by the tesseract OCR. Does anybody know, where are those tif files ? Th

Re: About the jpn.traindata

2011-05-17 Thread Dmitri Silaev
I think copyright issues are preventing the dev team from publishing these source files. However you can try to contact this forum's moderator directly - he probably can take decision to share. -- Dmitri On Tue, May 17, 2011 at 4:58 AM, Mostafa wrote: > Hi, > > I am interested to get all the

Re: mftraining produces "Missing font_properties"

2011-05-17 Thread zdenko podobny
On Tue, May 17, 2011 at 11:58 AM, Eyal wrote: > Quite a good guess, but I'm very disappointed to to say - I DID read the > documentation... > > And I even run the following command: > > *mftraining -F font_properties -U unicharset font1.tr* > > And I got results which don't show any error... : >

Re: About the jpn.traindata

2011-05-17 Thread Илья
IMHO alphabets can't be protected by copyright. -- Best regards, Ilia. В Втр, 17/05/2011 в 09:24 -0400, Dmitri Silaev пишет: > I think copyright issues are preventing the dev team from publishing > these source files. However you can try to contact this forum's > moderator directly - he probabl

Re: About the jpn.traindata

2011-05-17 Thread zdenko podobny
On Tue, May 17, 2011 at 5:01 PM, Илья wrote: > IMHO alphabets can't be protected by copyright. > > Mostafa did not asked for an alphabets. He asked for 'all the tif files that used for creating...' and content of tiff file (e.g. scanned books) could be protected by copyright. -- > Best regards

Re: About the jpn.traindata

2011-05-17 Thread Илья
He need for table that contains all supported alphabetics characters. Also, Parts of scanned books could not be protected by copyright. Can you give any contacts of "jpn.traindata" dev team? -- Best regards, Ilia. В Втр, 17/05/2011 в 18:24 +0200, zdenko podobny пишет: > > >

How to improve Tesseract digit ocr accuracy?

2011-05-17 Thread mw18888
How to improve Tesseract digit character accuracy? In testing the Tesseract 3.0, we see that there is a reasonable chance that the Tesseract makes mistake while recognizing a. 3 as 8 or 8 as 3 b. 5 as 6 or 6 as 5 c. 6 as 0 or 0 as 6 Can we configure (or training data) to improve the accuracy?