MCMC-Based Tracking and Identification of Leaders in Groups

Avishy Carmi (1) , Lyudmila Mihaylova (2) , François Septier (3, 4) , Sze Kim Pang (5) , Pini Gurfil (6) , Simon J. Godsill (7)

1 Nanyang Technological University [Singapour]
2 SCC - School of Computing and Communications [Lancaster]
3 LAGIS-SI
4 Institut TELECOM/TELECOM Lille1
5 DSO National Laboratories
6 Faculty of Aerospace Engineering
7 Signal Processing Group Dept of Engineering

We present a novel framework for identifying and track- ing dominant agents in groups. Our proposed approach relies on a causality detection scheme that is capable of ranking agents with respect to their contribution in shap- ing the system's collective behaviour based exclusively on the agents' observed trajectories. Further, the reasoning paradigm is made robust to multiple emissions and clutter by employing a class of recently introduced Markov chain Monte Carlo-based group tracking methods. Examples are provided that demonstrate the strong potential of the pro- posed scheme in identifying actual leaders in swarms of in- teracting agents and moving crowds.

Domains: Engineering Sciences [physics]; Signal and Image processing; Computer Science [cs]; Signal and Image Processing; Statistics [stat]; Applications [stat.AP]

Avishy Carmi, Lyudmila Mihaylova, François Septier, Sze Kim Pang, Pini Gurfil, et al.. MCMC-Based Tracking and Identification of Leaders in Groups. IEEE International Conference on Computer Vision Workshops (ICCV Workshops), Nov 2011, Barcelona, Spain. pp.112-119, ⟨10.1109/ICCVW.2011.6130232⟩. ⟨hal-00685921⟩