Tracking and Counting People in Complex Environment

Abstract

Tracking and counting multiple humans in complex situations is challenging. The difficulties are tackled with appropriate knowledge in the form of various models in our approach. Human motion is decomposed into its global motion and limb motion. Multiple human objects were segmented and their global motions were tracked in 3D using ellipsoid human shape models. An improved method aiming to estimate accurately the counts of people in a scene without constraining to individuals in complicated scene was proposed. Accurately estimate peoples which have an arbitrary shape or those due to dense crowds moving arbitrarily in a scene was discussed as well.