People Counting using Skeleton Graph and Tracking

Kheir-Eddine Aziz, Djamel Merad, Bernard Fertil, and Nicolas Thome


Head detection, Skeleton graph, Head pose estimation, Head tracking


In this paper, we describe a new process for people counting. This method is based on both the head detection using skeleton graph and the head tracking using particle filter. The shape of the head is modeled as the segment derived from the skeleton silhouette. Several results present the efficiency of the labeling human body parts, particularly its structural property for a head detection and to distinguish among persons in crowded environments. The proposed algorithm can track a head reliably in cases of temporal occlusions by dealing with multiple hypotheses for the pose. Each detected head in a skeleton silhouette is identified as independent state or partially occluded state, and during the tracking every state is updated. A proposed method has been tested with an experiment of counting the number of pedestrians passing through an indoor/outdoor transition area.

Important Links:

Go Back