125
click to vote
CVPR
2006
IEEE
15 years 9 months ago
2006
IEEE
We present an algorithm for jointly learning a consistent bidirectional generative-recognition model that combines top-down and bottom-up processing for monocular 3d human motion ...