Authors: D Rowe J Gonzàlez M Pedersoli J J Villanueva
Publish Date: 2009/04/03
Volume: 21, Issue: 2, Pages: 113-127
Abstract
This work develops a new architecture for multipletarget tracking in unconstrained dynamic scenes which consists of a detection level which feeds a twostage tracking system A remarkable characteristic of the system is its ability to track several targets while they group and split without using 3D information Thus special attention is given to the featureselection and appearancecomputation modules and to those modules involved in tracking through groups The system aims to work as a standalone application in complex and dynamic scenarios No apriori knowledge about either the scene or the targets based on a previous training period is used Hence the scenario is completely unknown beforehand Successful tracking has been demonstrated in wellknown databases of both indoor and outdoor scenarios Accurate and robust localisations have been yielded during longterm target merging and occlusions
Keywords: