Janez Perš, Faculty of Electrical Engineering, University of Ljubljana

Visual Sensor Networks in Theory and Practice

Abstract

Capabilities of embedded processors have been increasing at a rapid pace for a some time, reaching the point where even more complex computer algorithms can be implemented on embedded systems. Consequently, we have witnessed a migration of many computer vision tasks to the domain of embedded systems, and, at least conceptually, into the field of visual sensor networks (VSNs).

However, the the large-scale deployment of VSNs is hindered by many open problems, and in this talk, I will try to address three of them. I will present the conceptual solution for data-centric routing in VSNs, especially for typical surveillance scenario of detecting the appearance of the same object appearing throughout the very large network of cameras without a central server. Next, I will address the problem of performance evaluation in distributed surveillance scenarios. Finally, I will describe the challenges of designing the architecture for intelligent cameras, used as VSN nodes.

Short Biography

Janez Perš received the B.Sc., M.Sc., and the Ph.D. degrees in electrical engineering from the Faculty of Electrical Engineering, University of Ljubljana , Slovenia, in 1998, 2001, and 2004, respectively. He is currently an associate professor with the Machine Vision Laboratory, Faculty of Electrical Engineering, University of Ljubljana, where he teaches post-graduate course on Machine Vision. His current research interests include object tracking, human motion analysis, dynamic motion-based biometry, autonomous and distributed
systems, and machine vision applications in manufacturing and sport.

Simone Gasparini, Institut National Politechnique de Toulouse, France

Camera Tracking for Visual Effects in Movies

Abstract

In this talk I will present the on going work at our laboratory focusing on camera tracking for the previsualization of visual special effects during the shooting of movies. The work has been carried on in the context of the european funded project POPART (http://popartproject.eu/): the project aimed at bridging the gap between the production and the post production process in filmaking, bringing augmented reality directly on the set to pre-visualize the visual effects, so that the movie director can have a rough preview of the digital effects (3D rendering) that will be later added during the post-production process. I will described the original approach that has been implemented to combine quick and easy setup with a robust and precise camera tracking. 

Short Biography

Simone Gasparini is an Assistant Professor at Institut National Politechnique de Toulouse since 2013. He received the Laurea degree in computer engineering from Politecnico di Milano in 2003 with a thesis in the field of Robot Mapping. He then received the Ph.D. degree in Information Technology at Politecnico di Milano in 2007 with a thesis on 3D reconstruction from single images with omnidirectional cameras. He spent 4 years with the PERCEPTION group at INRIA Grenoble - Rhone-Alpes in Grenoble (FRA) where he mainly worked on geometrical aspects of computer vision, such as calibration,  real-time structure-from-motion and general camera models.