J. Storrs Hall, PhD. writes:
I'm intending to do lo-level vision on (one) 8800 and everything else on my
(dual) Clovertowns.
Do you have any particular architectures / algorithms you're working on?
Your
approach and mine sound like there could be valuable shared effort...
First I'm going to build a robot. While I do that, I'm going to learn how
to use the GPU hardware, read a lot, and figure out what to do next.
However, I'm definitely planning on starting with low level vision on the
8800 so we're certainly going in the same direction in that regard. So far
I'm capturing video from a firewire webcam using the CMU 1394 camera driver,
but haven't yet started doing much with the data except displayi it. It
should be possible to run hundreds of different convolutions on image data
in realtime so I'm planning to do that as a learning project.
I'm curious whether a clustering algorithm would automatically come up with
useful convolution kernels naturally simply by watching vast quantities of
image data (somebody must have tried that at some point), but it also isn't
too hard to hardcode a bunch of oriented edge detectors, endpoint detectors,
corner detectors, and whatnot. I have no idea where to go from there at
this point, but that's the fun of it.
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=231415&user_secret=fabd7936
