Your Perl-driven cyborgs can soon recognize objects

I've made some progress in getting ccv to compile under Windows. Now I can start with opening up the API and writing some XS for it. Soon, the below image can be created completely from within Perl instead of using a shell call out to the executable to find "interesting" points in the object and the scene and match them together.

out-sofa-thumb.png

The SIFT algorithm seems to be quite a useful tool to do feature recognition without any prior training. I hope to (ab)use this to recognize objects and how they move around in movies. Also, I imagine this can be used to automatically stitch images together.

3 Comments

Cool stuff! Keep us posted!

This guy Liu Liu is a genius.
The training step is a bit hard, comparable to tesseract.

Awesome! Any chance to get it working together with PDL?

Leave a comment

About Max Maischein

user-pic I'm the author of various CPAN modules. I'm also one of the admins of perlmonks.org.