Tablet PC Rorschach Inkblot Test
LPH writes "ESC Technologies just put up a funny series of images that they ran through the Tablet PC recognition software." Perhaps these tablets need a "doodle" setting.
← Back to Stories (view on slashdot.org)
The site's /.'ed so I haven't seen the pictures, but I tried out one of the tablet PC's at CompUSA yesterday and I was totally blown away by how good the handwriting recognition was. I scribbled down a few notes as I would on paper, and when I converted it to text it was perfect. I tried like 10 times and there was only 1 very understandable mistake (a humad transcribing what I had written probably would have thought the same thing.)
As far as "it needs a doodle setting", the apps that I used saved things as digital ink by default, and only converted to text when you specifically told it to.
Linking to a PHP forum right on the Slashdot main page isn't generally a good idea... :)
Thanks for the images.
:-)
:-)
For all the yappers. This link was a test to see if the servers could handle it. The test is a success because -- the servers couldn't handle the traffic !
OK. Back to networking 101
But seriously, the person who posted (aka ME) this wanted to see if PostNuke, MySQL, and Apache on a GNU/Linux server hosted by a company back east could handle the traffic. ESC Technologies didn't know that I submitted the link up there until last night. They were a bit concerned
And the company who submitted the story to whatisnew.com is inkwalker.com. So - go there and see what is all about !
LPH
Agreed. After finally getting to see the images from the slashdotted article, I can tell that what they did was:
a) doodle.
b) select the doodle
c) tell the handwriting engine to try to recognize their doodle as text.
d) laugh when it can't
Did you really expect to be able to draw a picture of a book and have it come back as the text "book?" What's next? Computer pictionary?
Hi,
What exactly would you expect by running a recognizer on such drawings? Garbage in, garbage out...
The only thing that seems weird is that is would seem possible for the recognizer to know that the approximation he found was really far fetched.
When I wrote a C# grafitti application a couple months back, if the gesture didn't match any letter close enough, the recognizer would not try to match it at all. So if you write a non-sense letter you don't get any result back.
Maybe the Tablet recognizer could have had something similar. But on the other hand the user is given a chance to fix whatever the Tablet recognized, so the current model seems fine.
The Tablet seems like a great product, if only I could get the one that I want (it's backordered everywhere in the US). I just wonder why isn't there more reviews and comparisons and testimonials all over the web (like there is for the iPod for example).
Are users happy with them? What do they use them for (browsing, reading divx, playing mp3,...)?
See you,
Dumky