Unicode character recognition by drawing

[Richmond]

That's a clever piece of work.

I threw: Ж , ѭ, ह and छ at it, and the only one it got wrong was the 
last [Devanagari 'ccha']; pretty good

(well, better than Meatloaf in "Two out of three ain't bad."); 
unfortunately as my non-Latin writing systems are
limited to about 6 (Old Cyrillic, Glagolitic, Hebrew, Devanagari, 
Grantha, Greek) I cannot remark how the thing
might do with, say, either Chinese characters or Japanese Kanji, but I 
doubt it would do as well.

Let's hope somebody who knows those systems "has a go" and reverts here.

But, clever or not, one wonders what ultimate value that has embedded in 
a webpage.

Of course, if one could "winkle the thing out" and fold it into a 
Livecode stack I'd be the first one to
be having "spontaneous cough, cough, coughs" as I'm fairly bonkers about 
kinky writing systems.

Thanks, Mark; that cheered me up after a rather "non" day!

Richmond.


On 12.07.2016 20:58, Mark Wieder wrote:
> Possibly of interest:
>
> http://shapecatcher.com/
>