on 27-05-2011
The presentation will be May 27th, at 10h00
Location: DCC - FCUP, sala 1
Rua Campo Alegre 1021.
Human Motion: Tracking and recognition of actions, emotions and interactions
J. K. Aggarwal
Department of Electrical and Computer Engineering
The University of Texas, Austin, Texas 78712
Humans have always been interested in motion: Mobiles hung over the crib fascinate young children. Zeno studied moving arrows to pose a paradox. Zeke is investigating the human brain devoted to the understanding of motion. Prof. Aggarwal’s interest in motion started with the study of motion of rigid planar objects and gradually progressed to the study of human motion. Understanding human motion is a diverse and complex subject that includes recognizing and tracking individual actions, interactions between people, and interactions between people and objects, from the actions and emotions of an isolated person to the actions and interactions of a crowd. Prof. Aggarwal’s talk will present an overview of the ongoing research in human motion recognition at The University of Texas at Austin. The issues considered in these problems will illustrate the richness and the difficulty associated with understanding human motion. The application of the above research to monitoring will also be discussed.
Short Bio:
J. K. Aggarwal has served on the faculty of The University of Texas at Austin College of Engineering since 1964. His research interests include computer vision, pattern recognition and image processing focusing on human motion. He is a Fellow of IEEE, IAPR and AAAS. More recently, he is the recipient of the 2004 K S FU prize of the International Association for Pattern Recognition, the 2005 Kirchmayer Graduate Teaching Award of the IEEE and the 2007 Okawa Prize of the Okawa Foundation of Japan. He is also a Life Fellow of IEEE and Golden Core member of IEEE Computer Society. He has authored or edited several books, chapters, conferences proceedings, and papers.
More Information..
on 27-05-2011
David Henriques (SQIG-IT).
May 27, 2011, Friday, 16h15m.
Abstract: Model checking has been extensively used to automatically check properties in large-scale systems. However, the classical techniques only allow for checking of qualitative properties, thus being inadequate for inherently probabilistic systems. Probabilistic model checking (PMC) extends the classical suite of techniques of classical model checking by allowing quantitative reasoning over probabilistic-nondeterministic systems which combine both probabilistic behaviour and nondeterministic choice. Abstraction, the main tool to deal with the ``state explosion'' problem in the classical setting, is not well developed in the probabilistic setting, limiting the applicability of PMC to little more than checking of reachability properties. Part of the reason for this limited success is that counterexamples in the probabilistic cases are much more complicated objects than their classical counterparts, making counterexample analysis much harder. In this seminar, we will present an abstraction-refinement loop for probabilistic systems based on an extension of the concept of may and must abstractions in the classical setting. May and must abstractions are respectively over and underestimations of the behaviours of a system. For a large class of properties, one of the abstractions preserves satisfaction and the other preserves non-satisfaction, eliminating the need for (expensive) counterexample analysis. Joint work with Anvesh Komuravelli and Edmund Clarke.
Room: 3.10, Mathematics
Support: SQIG/IT with support from FCT and FEDER, namely via the following projects:
* PTDC/MAT/68723/2006 KLog;
* POCI/MAT/55796/2004 QuantLog;
* POSC/EIA/55582/2004 Space-Time-Types
More Information..