Interactive marker-less tracking of human limbs

One of the hard problems that combines research in Computer Graphics and Computer Vision is tracking human limbs in real time without using any markers. This problem is known to be notoriously difficult and ill posed. In this paper we present a technique to track human limbs at near interactive rates without markers. This work builds on, modifies, and adds to the state-of-theart tracking techniques developed by previous researchers. We use a particle filtering tracking algorithm that uses 3d data derived from a modified visual hull algorithm. The particle filtering algorithm runs on a GPU and hence we are able to track human limbs at 4Hz with reasonable accuracy. This method has the potential to be applied to tracking the whole human body without markers.