You can try by training cascade detector by the different parts. You can apply human detection: http://www.magicandlove.com/blog/2011/08/26/people-detection-in-opencv-again/
and then apply the trained head and torso detectors on the resulting bounding box.
I'm making my people detector, and I wanna to run one last detector, that will verify the result. By that i wanna to reduce the percent false positives. I was thinking for some type of fitting of rectangle or something...
p.p. for example the car in the upper right corner
Asked: 2013-08-21 05:06:43 -0600
Seen: 6,437 times
Last updated: Aug 21 '13
your image seems to be broken ? only feet visible.