hi,
that sounds like a really cool project :)
I think for pygame it wouldn't be a good idea to depend on a library
as big as opencv. But it could be cool as an option.
Have a look on the pygame mailing list for a link to some code by Gary
Bishop who has made a ctypes based wrapper for using webcams with
opencv.
http://article.gmane.org/gmane.comp.python.pygame/13425/match=opencv
I use opencv myself, and quite like it, but for pygame on different
platforms it would be too much of a dependency. It would be very cool
for OLPC though :) Especially if opencv is installed by default.
Also note that pygame 1.8 has code for thresholding, differencing,
averaging, and bitmasks which are all good for webcam programs.
cheers,
On Sat, Apr 5, 2008 at 8:27 PM, Chris Hager <[EMAIL PROTECTED]> wrote:
> Hi all,
>
> On the pygame gsoc ideas page i saw the webcam project to integrate
> computer vision. I'm an enthusiastic pygame developer myself, mostly for the
> XO laptop. I am also soc mentor for OLPC and we have a very interesting
> OpenCV proposal.
>
> Both the student and I had previous contact with OpenCV and are very
> enthusiastic about the possibilities of this library (they are so quick, and
> run very well on the XO). Here is a link to Nirav's current experiments:
> http://eclecti.cc/olpc
>
> Here is a piece from the proposal draft:
>
>My proposal is thus for the creation of an interface layer between
>the v4l2 driver in the kernel and Sugar Activities. This interface
>would include functions to abstract and simplify basic and complex
>vision processes and algorithms. These would include colorspace
>conversions, thresholding, tracking connected components,
>convolving, motion tracking, and pattern recognition. The library
>itself could be from a well developed project like OpenCV to
>minimize code duplication. The next step is to build example
>Activities to demonstrate and document ways to make use of vision
>processing. The first is adding vision to an Activity like Colors!
>or Bounce to show how it can augment an existing experience. The
>second is creating a simple Activity from scratch to show how vision
>can work itself, such as xeyes with face tracking. Beyond that, with
>the toolkit, anyone interested could write Activities for motion
>based games, sign language recognition, a cartoon-based video chat,
>face detection and recognition, and so on.
>
>
> We would be delighted to discuss the possibilities of this project, also
> regarding pygame and what ideas you would have.
>
> Regards from Vienna,
> - Chris
>
> --
> \___/Chris Hager
> |___|GSoC '08 Mentor for OLPC -- http://wiki.laptop.org
> \___\ Elements Developer -- http://elements.linuxuser.at
>
>
>
>
>
