Bayesian Integration of a Discrete Choice Pedestrian Behavioral Model and Image Correlation Techniques for Automatic Multi Object Tracking (MP-P6)

Author(s) : Santiago Venegas - Martinez (Signal Processing Institute LTS-EPFL, Switzerland) Gianluca Antonini (Signal Processing Institute LTS-EPFL, Switzerland) Jean-Philippe Thiran (Signal Processing Institute LTS-EPFL, Switzerland) Michel Bierlaire (Operation Research EPFL, Switzerland)

Abstract : In this paper we deal with the multi-object tracking problem, with specific reference to the visual tracking of pedestrians, assuming worked out the pedestrian-detection step. We use a Bayesian framework to combine the visual information provided by a simple image correlation algorithm with a behavioral model ( discrete choice model ) for pedestrian dynamic, calibrated on real data. We aim to show how the combination of the image information with a model of pedestrian behavior can provide appreciable results in real and complex scenarios.

Menu