[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [pygame] GSoC: Vision Processing and Webcam Support in Pygame



Thanks René,

I'll be blogging about the project at http://eclecti.cc/olpc

Thanks for the svn offer.  I'm going to be in a country with spotty
internet access for most of the summer, so I'm going to be working on
a local svn or git repository.  It could be useful for me to mirror my
work on pygame's svn though.

I somehow missed pygame.transform completely.  Based on the functions
in it, it seems to be the best place for the vision functions I want
to add.

Nirav Patel

On Sun, May 11, 2008 at 10:04 PM, René Dudfield <renesd@xxxxxxxxx> wrote:
> Hello,
>
> congratulations on being accepted :)  It sounds like a fun and
> interesting project.
>
> Do you have a blog you will be using to discuss your project?
> I'd like to add it with the other GSOC project blogs...
> http://www.pygame.org/wiki/rsslinks
>
> If you like I can set you up with a branch on the pygame subversion to
> do your work on there?
>
> pygame has thresholding (see pygame.transform).
>
> Marcus (nick: __raz__) is working on colorspace conversions as part of
> his pixel array work.
>
>
> If you have any questions about things... like what API to use, or
> where things should go in pygame, please feel free to ask on the
> mailing list.
>
>
> cheers,
>
>
>
>
> On Sat, May 10, 2008 at 11:26 AM, Nirav Patel <olpc@xxxxxxxxxxxxxx> wrote:
>> Hello all,
>> I'm participating in Google Summer of Code with One Laptop Per Child,
>> and my project is to write a computer vision library that works on the
>> OLPC XO.  I plan on doing this by incorporating webcam support and
>> some vision functions into Pygame.
>>
>> In the next few days, I will start writing a v4l2 interface that
>> initializes a v4l2 camera and loads a frame of video as a surfarray.
>> This will basically be a python wrapper around something like
>> http://www.linuxtv.org/downloads/video4linux/API/V4L2_API/spec/capture-example.html
>>
>> The next step will be writing vision functions.  At minimum, I would
>> like to do thresholding, colorspace conversion, histograms,
>> convolution, and centroids of connected components.  Basically, things
>> that would be useful for gaming purposes.  It would also be useful to
>> have some kind of motion detection using multiple surfarrays and an
>> algorithm like the Lucas-Kanade method.  Most things past that, like
>> Haar feature detection for face detection, may be getting too heavy
>> for inclusion in Pygame.
>>
>> Any suggestions, questions, or comments would be greatly appreciated.
>>
>> Thanks,
>> Nirav Patel
>>
>