Nonparametric Bayesian Methods and the Dependent Pitman-Yor Process for Modeling Evolution in Multiple Object Tracking

In this paper, we propose a family of dependent Pitman-Yor (DPY) processes to model the state prior for multiple object tracking. This process is shown to be more flexible and a better match than the dependent Dirichlet process in tracking a time-varying number of objects. The DPY model directly incorporates learning multiple parameters from correlated information. Integrated with a Dirichlet process mixture model, the overall approach estimates time dependent object cardinality, provides object labeling, and identifies object associated measurements. Using Markov chain Monte Carlo sampling methods, the performance of the DPY based approach is demonstrated and compared to the labeled multi Bernoulli tracker.