Aedict Ultimate, Kanji Recognition

[Pirin]

Hello,

first of all, im glad, that i can support you with Aedict Ultimate and give something back for a very helpful application. 🖖🏼 Really glad, i can have an app like that. Thanks a lot for your work.

One thing though, still doesnt work as it should. The Kanji Recognition is so rigid, seems to not recognize the position and relations between strokes, it rarely works. The Microsoft IME is way better with that. Please, this needs a fix, cause its a really important feature.

Second bug, lately the Aedict OCR reduces the image resolution and then can't read the kanji properly

[Martin Vysny]

Hi Pirin,

  thank you so much for your kind support, it really means a lot and
makes world of difference :)

Sadly, Microsoft IME is probably using some kind of really neat
algorithm which is impossible to obtain, and would require a lot of
effort to replicate :( I tried to find a good Java-based stroke
recognizer; the current one which Aedict uses is the best I could find.
However, it might not hurt to try evaluate those libraries again. Could
you please open a bug report at https://github.com/mvysny/aedict/issues
regarding this?

Regarding the Aedict OCR: I'm sorry to hear that. Could you please open
a bug report at https://github.com/mvysny/aedict/issues and attach the
image itself and maybe a screenshot of your phone's screen there?

Thank you,
Best regards,
Martin

[Martin Vysny]

Hi Pirin,

  just a question: there are two KanjiPads in Aedict, the better one
gets activated with the purchase of Aedict Online. You can tell whether
the "better one" is activated easily: if there is a "+-2 strokes"
checkbox then the better kanjipad is in use. Could you perhaps verify
that the better kanjipad is in use for you, and perhaps compare it to
the Microsoft IME?

Thanks,
Best,
Martin