From Handcrafted to End-to-End Learning, and Back, seminar by Laura Leal-Taixé
On Friday, December 2nd, 2022, at 2:15 p.m., Laura Leal-Taixé (NVIDIA, Technical University of Munich) gives a talk on “From Handcrafted to End-to-End Learning, and Back: a Journey for Multi-Object Tracking”. The seminar is part of the cycle GENDER UNBALANCED AI and is held in the Sala Eventi of the Modena Technopole (Building 52) of the “Enzo Ferrari” Department of Engineering, and online.
Abstract
The challenging task of multi-object tracking (MOT) requires simultaneous reasoning about track initialization, identity, and spatiotemporal trajectories. This problem has been traditionally addressed with the tracking-by-detection paradigm, but recent research has focused on end-to-end learning paradigms such as tracking-by-regression or tracking-by-attention. In this talk I will discuss all the paradigm shifts, only to circle back right where we started: tracking-by-detection. Can this paradigm be state-of-the-art?
More information is available in the seminar poster.