I will be talking about a unified three-layer hierarchical approach for solving tracking problems in a multiple non-overlapping cameras setting is proposed. Given a video and a set of detections (obtained by any person detector), the approach first solves the within-camera tracking employing the first two layers of the framework and, then, in the third layer, it solves across-camera tracking by associating tracks of the same person in all cameras in a simultaneous fashion. The tracking problem is formulated as finding fast-constrained dominant sets from a graph. The approach is based on a novel Fast-Constrained Dominant Set Clustering (FCDSC), which is an order of magnitude faster than constrained dominant sets clustering technique. FCDSC is employed to solve both within- and across-camera tracking tasks. That is, given a constraint set and a graph, FCDSC generates cluster (or clique), which forms a compact and coherent set that contains whole or part of the constraint set. The approach is based on a parametrized family of quadratic programs that generalizes the standard quadratic optimization problem. In addition to having a unified framework that simultaneously solves within- and across-camera tracking, the third layer helps to link broken tracks of the same person occurring during within-camera tracking. We have tested this approach on a very large and challenging dataset (namely, MOTchallenge DukeMTMC) and show that the proposed framework outperforms the current state of the art.
Even though the main focus of this paper is on multi-target tracking in non-overlapping cameras, proposed approach can also be applied to solve video-based person re-identication problem. We show that when the re-identication problem is formulated as a clustering problem, FCDSC can be used in conjunction with state-of-the-art video-based re-identication algorithms, to increase their already good performances. Our experiments demonstrate the general applicability of the proposed framework for non-overlapping across-camera tracking and person re-identification tasks.
Eyasu received the BSc degree in Electrical Engineering from Jimma University in 2007, he then worked at Ethio Telecom for 4 years till he joined CaFoscari University (October 2011) where he got his MSc in Computer Science in June 2013. September 2013, he won a 1 year research fellow to work on Adversarial Learning at Pattern Recognition and Application lab of University of Cagliari. Since September 2014 he is a PhD student of CaFoscari University under the supervision of prof. Pelillo. Working towards his Ph.D. he is trying to solve different computer vision and pattern recognition problems using theories and mathematical tools inherited from graph theory, optimization theory and game theory. From April 2016 to September 2017, Eyasu was working, as a research assistant, at Center for Research in Computer Vision at University of Central Florida under the supervision of Dr.Mubarak Shah. Currently, he is a researcher at Qualcomm Technologies Inc. His research interests are in the areas of Computer Vision, Pattern Recognition, Machine Learning, Graph theory and Game theory.