In this paper, we present a comparison of four novel algorithms

that have been applied to the tracking of people in

an indoor scenario. Tracking is carried out in 3D or 2D

(ground plane) to provide position information for a variety

of surveillance, HCI or meeting-support services. The

algorithms, based on background subtraction, face detection,

particle filter feature-matching and edge alignment

of a cylindrical model are described and comparative results

presented using independent test data produced and

ground-truthed for the EU CHIL project.