I am exhausted figuring out how the user-words and user-patterns work?
I did over 10 different experiments.
the result never matched the word I put in user-words.
What are the possible reasons?
Thanks again in advance.
--
You received this message because you are subscribed to the Google
I do not need a zillion fonts or images though,
just train some numbers.
I read through all the posts.
didn't see much cons for upgrading to 3.03 though.
any hints?
thanks in advance.
--
You received this message because you are subscribed to the Google Groups
tesseract-ocr group.
To
https://lh4.googleusercontent.com/-Cop8Qo5A9VQ/U8d3mOfAyWI/Aos/Lu9gT01DbKo/s1600/bounding+box+without+actual+letters+touching.png
The Ray's tutorial said the bounding box overlaps.
so when I modify the box inside JTessbox,
do I keep the overlapping boxes,
or
make the boxes non
https://lh5.googleusercontent.com/-yFFAJvw4F1U/U8d3NPEBjcI/Aok/G8IIxiz5aLA/s1600/selecting+.png
seems not only does eng.cube.freq-words work.
it is depended on other factors. too
--
You received this message because you are subscribed to the Google Groups
tesseract-ocr group.
To
I am going through Ray Smith's tutorial, and don't get it?
anyone sheds some light on it? thank you.
https://lh4.googleusercontent.com/-zJbwuVTCRg4/U8dqBCSoIEI/AoU/SSWEXrP9LwY/s1600/no+silly+%2B1%3A-1.png
On Monday, May 23, 2011 10:50:31 AM UTC+9:30, Mostafa wrote:
Dear Mr. Smith
Hope you passing a lovely day.
I had post a FAQ about the jpn.traineddata which thread is as below:
On Wed, Jul 16, 2014 at 11:17:00PM -0700, Jing JC wrote:
I am going through Ray Smith's tutorial, and don't get it?
He means that as the co-ordinate system uses bottom left as the
origin, you will never get a minus number co-ordinate (as you could
if the origin was elsewhere).
--
You
On Thu, Jul 17, 2014 at 12:14:43AM -0700, Jing JC wrote:
The Ray's tutorial said the bounding box overlaps.
so when I modify the box inside JTessbox,
do I keep the overlapping boxes,
or
make the boxes non touching.
That's interesting, actually; I didn't realise Tesseract did
outlining
Citing from the Wiki
(https://code.google.com/p/tesseract-ocr/wiki/TrainingTesseract3#Old_Manual_method):
It is sometimes important to space out the text a bit when printing, so up
the inter-character and inter-line spacing in your word processor. Not
spacing text out sufficiently will cause
[root@centos57 AdaleMono]# shapeclustering -F font_properties -U unicharset
eng.matrx60x40.exp0.tr
Reading eng.matrx60x40.exp0.tr ...
Error: Unable to open eng.matrx60x40.exp0.tr!
signal_termination_handler:Error:Signal_termination_handler called:Code 3000
Segmentation fault
yep agree.
didn't end up having overlapping during training stage yet.
On Thursday, 17 July 2014 11:32:59 UTC-7, Paul wrote:
Citing from the Wiki (
https://code.google.com/p/tesseract-ocr/wiki/TrainingTesseract3#Old_Manual_method
):
It is sometimes important to space out the text a bit
yep yep.
it happened during the bounding boxes I generated myself.
not happened to the .box during the training step yet.
On Thursday, 17 July 2014 10:17:39 UTC-7, Nick White wrote:
On Thu, Jul 17, 2014 at 12:14:43AM -0700, Jing JC wrote:
The Ray's tutorial said the bounding box
I followed the naming convention though:
tesseract [lang].[fontname].exp[num].tif [lang].[fontname].exp[num] nobatch
box.train
while my font name contained number:
my font name is matrx60x40,
I changed to the name arial,and it worked.
weird.
I tried replacing the font name to
Hello Nick
It is great that you are motivated to make a documentation and that you
answer the questions in the forum.
Nevertheless I read a post from Ray where he says that he receives millions
of emails and the last thing he likes to do is writing long texts (email
responses or
14 matches
Mail list logo
