Abstract
We develop an interactive likelihood (ILH) for sequential Monte Carlo (SMC) methods for image-based multiple target tracking applications. The purpose of the ILH is to improve tracking accuracy by reducing the need for data association. In addition, we integrate a recently developed deep neural network for pedestrian detection along with the ILH with a multi-Bernoulli filter. We evaluate the performance of the multi-Bernoulli filter with the ILH and the pedestrian detector in a number of publicly available datasets (2003 PETS INMOVE, Australian Rules Football League (AFL) and TUD-Stadtmitte) using standard, well-known multi-target tracking metrics (optimal sub-pattern assignment (OSPA) and classification of events, activities and relationships for multi-object trackers (CLEAR MOT)). In all datasets, the ILH term increases the tracking accuracy of the multi-Bernoulli filter.
Highlights
Multi-target tracking (MTT) is a well-researched problem, with a history going back over 50 years [1]; it remains an open research problem [2,3]
We evaluate the performance of the multi-Bernoulli filter with and without the interactive likelihood in a number of publicly available datasets and obtain quantitative results using standard, well-known metrics
2003 PETS INMOVE: In this dataset, the performance of the multi-Bernoulli filter without (MBF) the interactive likelihood (ILH), with the ILH (MBFILH), an implementation of the multiple hypothesis tracking (MHT) method [67], the multi-Bernoulli filter without the ILH and with a fixed target size (MBF FS), and the multi-Bernoulli filter with the ILH with a fixed target size (MBFILH FS) is evaluated; the HSV-based likelihood function in Equation (8) is used for all RFS
Summary
Multi-target tracking (MTT) is a well-researched problem, with a history going back over 50 years [1]; it remains an open research problem [2,3]. It has many applications, including aviation [4] and air traffic control [5], ballistic missile defense [6,7], smart surveillance [8,9], robotics [10] and autonomous vehicles [11,12,13,14,15]. The goal of MTT is to simultaneously estimate both the number of targets and their states (position, size, velocity, etc.) through time [16] This can be a difficult task for a number of different reasons; to name just a select few of these challenges:. Tracks are defined by Ristic et al [16] as a “labeled temporal sequence of state estimates associated with the same target”
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
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
