Videos for Ballet and Sign Language (sl) datasets

In all videos, red corresponds to the ground truth, green corresponds to our
method. Where applicable, blue shows the output of a linear dynamical system.

<dataset>_dyn.avi - Show tracking results of full model, mixture of GPs and the
dynamical pose filtering algorithm.

<dataset>_nodyn.avi - Show just the mixture of GPs appearance model.

<dataset>_dyn_vs_nodyn.avi - Compares the dynamics introduced in this paper
compared to a linear dynamical system.

