Abstract
A large-scale multi-object tracker based on the generalised labeled multi-Bernoulli (GLMB) filter is proposed. The algorithm is capable of tracking a very large, unknown and time-varying number of objects simultaneously, in the presence of a high number of false alarms, as well as missed detections and measurement origin uncertainty due to closely spaced objects. The algorithm is demonstrated on a simulated tracking scenario, where the peak number objects appearing simultaneously exceeds one million. Additionally, we introduce a new method of applying the optimal sub-pattern assignment (OSPA) metric to determine a meaningful distance between two sets of tracks. We also develop an efficient strategy for its exact computation in large-scale scenarios to evaluate the performance of the proposed tracker.
Highlights
M ULTI-OBJECT tracking is a problem with a wide variety of applications across diverse disciplines, and numerous effective solutions have been developed in recent decades [1]– [3]
A few notable examples are: (i) space situational awareness, which requires tracking thousands of satellites and millions of debris objects [4]–[6]; (ii) wide area surveillance, requires tracking hundreds of thousands of objects over time, including vehicles and people in crowded environments [7]–[9]; (iii) cell biology, where tracking the motion of large numbers of cells is critical to understanding their behaviour in living tissues [10], [11]; (iv) wildlife biology, where tracking large animal populations is needed to study the behaviour of wildlife in their natural habitats [16]
To evaluate the performance of the proposed tracker on a scenario involving an unknown and time-varying number of objects with a peak in excess of one million, we developed a scalable procedure for exact computation of the optimal sub-pattern assignment (OSPA)(2) metric
Summary
M ULTI-OBJECT tracking is a problem with a wide variety of applications across diverse disciplines, and numerous effective solutions have been developed in recent decades [1]– [3]. The common goal of multi-object tracking is to estimate the trajectories of an unknown and time-varying number of objects, using sensor measurements corrupted by phenomena including observation noise, false alarms, missed detections, and data association uncertainty. The combination of these effects gives rise to a highly demanding computational task, with complexity that grows exponentially as the number of objects/measurements increases. The generalised labelled multi-Bernoulli (GLMB) filter is an algorithm that is designed to provide estimates of object trajectories by modeling the multi-object state as a labeled random finite set (RFS). This is expressed mathematically by defining a distinct label indicator function
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
