Bayesian Inference
	Build models of human form and motion.  Learn priors over model parameters:    p(model) Exploit cues in the images.  Learn likelihood models:    p(image cue | model) Represent the posterior distribution    p(model | cue)       p(cue | model) p(model)