Hello Nathan
I'm interested in hearing more about your project, since it is
closely related to what I do consulting work in: information extraction from
OCR'ed document images, including phone camera images. I won't say to not
use Ocropus, but I would like to understand the application better to know
why you might want to use Ocropus (since it can be a headache to use).
Thomas L. Packer
~~~~~~~~~~~~~~~~~~~~
-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf
Of Nathan K
Sent: Wednesday, July 13, 2011 2:58 PM
To: ocropus
Subject: [ocropus] Ocropus Mobile, OpenFST for custom application, Training
resources
Hi All,
Thanks for all the great work in producing Ocropus! I'm hoping it can
help me digitize some content on mobile devices.
#Q1 - Ocropus Engine on iPhone and/or Andoid
I'm particularly interested in hearing members opinions about the best
way to get ocropus running on iOS and/or android. I'm largely
interested in using a trained 'lanuage' model to make predictions on
the device. Training will occur offline and off device. I know a few
people have got tesserect running on mobile devices, but I'm unable to
find much information on doing the same with Ocropus. I'd greatly
appreciate a slice of your collective wisdom in an effort to avoid
wasting days taking the wrong path.
Would it be easier to just prototype the algorithm using the scripts,
then grab the specific c++ code of interest and include it directly in
my application. Or best to compile as a static/dynamic library?
#Q2 - OpenFST for custom application? - Please bare with me this is a
new tool to me.
Also, I'm interested in using the probabilistic language model.
Currently I'm inexperienced with the particular library included. In
my application I the text I'm processing is found on an invoice and
follows the structure:
<Company name> <Item Name> <Size> \t\t <price>
...
...
Would training a application specific language model in OpenFST be
able to improve results for this application? There is no field
devider in the document, so I'm looking for a method to automatically
restrict symbol probabilities and make corrections when the highest
ranking returned OCR symbol does not fit its context.
#Q3 - Training Resources
I've found some references to a training course. However have been
unable to login when accessing it at
https://sites.google.com/a/iupr.com/ocropus-course/about
Anyone know how I might gain access to any extra documentation to
steepen my learning curve with Ocropus?
Note: I'm still unable to get the python bindings to work cleanly on
OSX. If anyone has achieved this I'd appreciate hearing from you as to
the steps involved. Cheers.
--
You received this message because you are subscribed to the Google Groups
"ocropus" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/ocropus?hl=en.
--
You received this message because you are subscribed to the Google Groups
"ocropus" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/ocropus?hl=en.
