The paper presents the development of a real time tracking system, named Watch Out, that is able to efficiently run on an Nvidia Jetson board mounted on a UAV (Unmanned Aerial Vehicle). The approach to long term video tracking implemented in Watch Out is named Best Structured Tracker (BST): a set of local trackers independently tracks patches of the original target in an online learning manner, while an outlier detection procedure filters out the less meaningful ones, and a resampling procedure allows to correctly reinitialise the trackers that have been filtered out. Performance of the tracking algorithm has been verified both on VOT2016 challenge datasets and in real situations using an Nvidia Jetson board mounted on a drone. Results show that the proposed system can track almost every possible target in real time.
Watch Out: Embedded Video Tracking with BST for Unmanned Aerial Vehicles
Francesco Battistone;Alfredo Petrosino;Vincenzo Santopietro
2018-01-01
Abstract
The paper presents the development of a real time tracking system, named Watch Out, that is able to efficiently run on an Nvidia Jetson board mounted on a UAV (Unmanned Aerial Vehicle). The approach to long term video tracking implemented in Watch Out is named Best Structured Tracker (BST): a set of local trackers independently tracks patches of the original target in an online learning manner, while an outlier detection procedure filters out the less meaningful ones, and a resampling procedure allows to correctly reinitialise the trackers that have been filtered out. Performance of the tracking algorithm has been verified both on VOT2016 challenge datasets and in real situations using an Nvidia Jetson board mounted on a drone. Results show that the proposed system can track almost every possible target in real time.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
