Well... now for some simple object recognition...
The program flow:
a) take an image; take a subsection as the reference object - calculate and store the SIFT descriptors for this image
b) rotate DRC by 180 degrees.
c) take an image
d) calc SIFT descriptors and look for matches with the reference
e) if not many matches - rotate a little and go back to c)
f) else: WooHoo! Lots of matches.... do a spin around in joy
And there's a little video... http://www.youtube.com/watch?v=WraE9UHzFAc
Next... DRC has to start a proper search... a bit more than rotating on the spot... :o)
https://www.youtube.com/watch?v=WraE9UHzFAc