I've been mulling about doing something like that for blue jays and peanuts to ration out their allotment in a sustainable manner. A family of blue jays (or most birds in the Crow [Corvidae] family) will rapidly hoard whatever you give them and finish off a large pile of peanuts within hours. Unlike other birds that also visit the pile of peanuts I leave out, they have no intentions of eating most of those peanuts as they take them and just stash them for later (similar to a squirrel).

The other problem is allocating peanuts to other birds (chickadees, nuthatches, cardinals, etc) while not giving too many to the blue jays. However, there's a project I ran across from CalTech that can determine the type of bird in front of a camera via computer vision[1][2][3]. Not sure why the creator of the squirrel project used a camera though when a cheaper, more robust sensor would work unless he was eventually thinking of extending its usage for more than simple motion detection.

[1] http://vision.caltech.edu/visipedia/20q.html

[2] http://vision.caltech.edu/visipedia/ipadapp.html

[3] https://github.com/welinder/cubam

Awesome work, that squirrel looks happy!

If anyone else has used motion on a Raspberry Pi, have you have any issues staying connected to wi-fi? I was running motion on my Raspberry Pi, trying to do something similar, and found that within 24 hours the wifi module would disconnect and stay down. I would have to reboot.

Without having motion running, I can stay connected to wifi for weeks at a time with the same pi/router/network. I haven't found a solution for this yet.

Cool :)

I wonder how hard it'd be to do 'squirrel recognition' with the camera