The seminar is held in occasion of prof. Garcia-Fernandez's visit to DINFO, within the collaboration with the University of Liverpool, financed by the Royal Society.
Date and time: monday, Jan 22, 3pm, room 173, S. Marta
via Zoom: https://liverpool-ac-uk.zoom.us/j/99410769899?pwd=Wm9Oc0dPSE5KTUZlMmZOa1RsNUFoZz09
Meeting ID: 994 1076 9899
Passcode: y$GA$E5E
Abstract: This talk will serve as an introduction to Bayesian multi-target tracking (MTT) systems. In MTT, there is an unknown number of objects that appear, move and disappear from a scene of interest, and we aim to estimate their states based on noisy measurements. This type of system with an unknown number of objects can be usually modelled via random finite sets (RFSs). For the standard dynamic and measurement models, the posterior density, which contains all information on the current set of targets, is given by the Poisson multi-Bernoulli mixture (PMBM) filter. The talk will explain the RFS and PMBM fundamentals, explain how Bayesian MTT can be combined with deep learning video detectors, show videos of applications to autonomous vehicle perception, and point out to available PMBM code online.