Bayesian Nonparametric View to Spawning

TL;DR

This paper introduces a Bayesian nonparametric framework for multi-object tracking that effectively handles spawning events, where measurements may originate from multiple objects, and the number of objects is unknown, using MCMC sampling.

Contribution

It presents a novel Bayesian nonparametric model for tracking with spawning, allowing for unknown object counts and complex measurement associations, with a tractable inference method.

Findings

Outperforms existing methods in scenarios with spawning events

Demonstrates the effectiveness of nonparametric modeling in multi-object tracking

Provides a tractable MCMC approach for posterior sampling

Abstract

In tracking multiple objects, it is often assumed that each observation (measurement) is originated from one and only one object. However, we may encounter a situation that each measurement may or may not be associated with multiple objects at each time step --spawning. Therefore, the association of each measurement to multiple objects is a crucial task to perform in order to track multiple objects with birth and death. In this paper, we introduce a novel Bayesian nonparametric approach that models a scenario where each observation may be drawn from an unknown number of objects for which it provides a tractable Markov chain Monte Carlo (MCMC) approach to sample from the posterior distribution. The number of objects at each time step, itself, is also assumed to be unknown. We, then, show through experiments the advantage of nonparametric modeling to scenarios with spawning events. Our experiment results also demonstrate the advantages of our framework over the existing methods.