skritter like took with tegaki
Emmanuel Oga
Roger Braun
2011/12/4 Emmanuel Oga <emman...@gmail.com>:
> Hi, I'm very new to this project and I'm wondering if there is a tool using
> tegaki to learn to write the kanji characters, something like scritter [1].
> If not, how hard would it be to write a tool like that? Thanks.
That looks pretty nice and polished, but I think it's a bit
expensive... It's hard to guess how much work it would be to write
something like this. IIRC, Tegaki does not do stroke level
recognition, so a showing recognizing a single stroke as written
correctly is not possible. Anyway, recognizing correct writing is not
actually the goal of a good handwriting recognition program, as it
should also recognize popular misswritings or shorthand / cursive
characters.
A program that would recognize the strokes should not be too hard,
though. You could take the data from http://kanjivg.tagaini.net/ and
use these to let the learner write stroke by stroke. Then compare
whatever they drew to the KanjiVG data, and count it as right if it's
in the right direction and not too different looking. Are you
interested in developing such a tool?
--
Roger Braun
rogerbraun.net | humoralpathologie.de
Emmanuel Oga
Roger Braun
2011/12/7 Emmanuel Oga <emman...@gmail.com>:
> Ah! now that you mention that... it makes sense. When recognizing day to day
> writing, you do not strive to recognize 100% correctness, but to make sense
> of real hand writing with all its mistakes.
Exactly.
>
> Yeah, I think it is a nice project to work on. I'll take a look at the
> kanjiVG data and see if I can make something out of it. Are you affiliated
> to kangiVG in any way or have experience using it?
I am sort of affiliated. I work on wadoku.eu, a front end for
WaDokuJT, which is a sister project of KanjiVG. Also, Ulrich Apel, the
creator of KanjiVG, is a coworker of mine.
--
Roger Braun