1
Programming using the Ultra Hal Brain Editor / Hal Web Cam vision.
« on: April 16, 2009, 10:47:29 pm »
I have been looking at this for a while, and have some thoughts and would like some feed back.
Most of the facial recognition posts here talk about giving Hal cam vision. I don't think that is a good idea.
Mostly because there are several very good Facial Recognition programs out there, and openCV is going to make progress fly here in a little bit. If FR is encoded into Hal, we are going to miss major leaps in FR just waiting for the next release of Hal to come out, instead of just compiling the latest distro.
My idea is to take an openCV program and use it, now.
The program just needs to be self contained where it can recognize faces and monitor emotion. Here is a plain example:
http://www.youtube.com/watch?v=V7UdYzCMKvw&feature=related
All you need to do is take the output from the program, lets's call it (happy,sad,angry,asleep), and send it to a file 2-3 times a second.
Havethe program keep a running tab on the emotions and where they are going, so your output is
---------------
User: David
Emotion: Happy+
---------------
This says I am the person, and my emotion is happy, and just for this I am using
= emotion stable
- emotion getting less (I.E. becoming less angry)
+ emotion getting more (I.E. becoming happier)
Then all you need to do is make a pluging to read the table before he calculated a sentence, and take in that he is still talking to David,
and David has getting happier as the conversation has progressed.
If it was done like this, it could be started right now, and would be just as useful or more so than having "vision" incorporated into Hal.
Thoughts?
Most of the facial recognition posts here talk about giving Hal cam vision. I don't think that is a good idea.
Mostly because there are several very good Facial Recognition programs out there, and openCV is going to make progress fly here in a little bit. If FR is encoded into Hal, we are going to miss major leaps in FR just waiting for the next release of Hal to come out, instead of just compiling the latest distro.
My idea is to take an openCV program and use it, now.
The program just needs to be self contained where it can recognize faces and monitor emotion. Here is a plain example:
http://www.youtube.com/watch?v=V7UdYzCMKvw&feature=related
All you need to do is take the output from the program, lets's call it (happy,sad,angry,asleep), and send it to a file 2-3 times a second.
Havethe program keep a running tab on the emotions and where they are going, so your output is
---------------
User: David
Emotion: Happy+
---------------
This says I am the person, and my emotion is happy, and just for this I am using
= emotion stable
- emotion getting less (I.E. becoming less angry)
+ emotion getting more (I.E. becoming happier)
Then all you need to do is make a pluging to read the table before he calculated a sentence, and take in that he is still talking to David,
and David has getting happier as the conversation has progressed.
If it was done like this, it could be started right now, and would be just as useful or more so than having "vision" incorporated into Hal.
Thoughts?