Statistical video analysis for crowds counting
Antonio Albiol, Alberto Albiol, Julia Silla · 2009
This paper presents an approach to count the number of people that enters or leaves metro trains. This is a challenging scenario where usually people crowd around the train doors, and therefore it is not possible a direct approach that segments and counts individuals. The proposed technique is based on a statistical analysis of the flow obtained from the motion vectors at corner points. The method has been evaluated on metro sequences showing an average error below 5% in real operational conditions.