sakarika / kanjitomo-ocr Goto Github PK

Gradle should be used for dependencies:

compile "org.imgscalr:imgscalr-lib:4.2-release"
compile "com.esotericsoftware:kryo:5.0.0-RC3"

Any suggestions on how to adjust text size in KanjiTomo? It is extremely tiny on my 4K monitor, in Windows 10. Looking around for a parameter I can maybe change, but no luck so far..

I've noticed that the OCR often has difficulty recognizing 小. It's typically like this with it only recognizing the strokes on the edges.

I gotta say, even though it hasn't been updated in a while and may never be, kanjitomo is still better than a lot of other options I've seen online even recently. Thanks for making this.

I'd like to make a modification to the original GUI for allowing direct adding of words to Anki. Are you willing to release that source code?

Obviously I can rewrite another GUI, but the less work the better :)

Due to android java not having some imageio classes, the project cannot be run on mobile. Is there a fork or mobile version?

I'm getting frequent mismatches between the output of this library and the output of Kanjitomo proper, sometimes quite significant. What might be causing this? Is this an outdated version?

Here's an example:

Kanjitomo correctly makes 想 the top pick for the second character, but for this library it doesn't appear anywhere in the top results.

(I don't think this matters, but I'm using it from Python via PyJnius.)

Cannot recognize drama subtitle in color. Cannot recognize full line in black white, it can only recognize several characters each time?
[email protected]

Hi, I have just a little question since I'm working on a similar program to KanjiTomo and am curious: how does KT pick the area close to the pointer to then send to OCR? It doesn't constantly OCR everything near the pointer to see if there's any character match, does it?

Loka, you can use this to translate the words from Japanese that you don't understand. They are written on a hoodie that you purchased with a picture of the anime "Avatar the Last Airbender". While you are pretty sure that kanji says "Avatar the Last Airbender", it would be good to confirm that the sweater does not have a rude message written instead. @lokafsenna

Could you please assist with updating dictionary? I see some .cache files inside KanjiTomoOCR.jar but have no idea how to produce them.

when i use the OCR on other windows, it usually pops up a black box over the characters it recognizes

from a screenshot on the page i assume its supposed to only be red boxes, but for me on linuxmint mate it shows black boxes almost always

sometimes it shows boxes within boxes but still obscures the original text

i looked for settings related to either disabling or customizing the ocr overlay window thing but couldn't find any

the main inconvenience of this is i can't confirm the OCR was accurate because the original character is hidden, if i click the window it goes away, but also removes the match from kanjitomo's window

heres an example
occasionally the black box doesn't appear at all, i think the window exists but is not in the foreground so not seen
so alot of inconsistency in its behavior