CALL FOR PAPER
(download PDF version)

Real time surveillance, environment monitoring, ambient intelligence, people behaviour and people interaction analysis are now key-applications of computer vision and sensor network systems. Multimedia paradigms, computer vision and pattern recognition techniques can be now integrated to feasible solutions of multi-source data processing.

The technologies of optical (fixed , PTZ, omni-directional cameras) and non optical (thermal, sound, biological,...) sensors, the high-speed, wired/wireless networks and of computing systems are mature enough to allow a serious testing of theories and models of real time multimedia knowledge extraction from distributed data.

The VSSN Workshop 2005 is the 3rd organized edition in conjunction with ACM Conference of Multimedia. Following the strong interest of the past editions, the workshop will bring together researchers, developers and practitioners from academia and industry to discuss of both theoretical aspects and practical implementations of new generations of video surveillance and sensor networks and novel multimedia applications in the field.

The workshop will cover these two aspects in separate sections:

1. theoretical aspects of video surveillance and sensor networks: new algorithms and models of computer vision and pattern recognition; new paradigms for managing surveillance multimedia data, new architectures of cameras and sensors, novel coordination and communication aspects.
2. innovative project implementation and applications: new civil and military applications, indoor our outdoor surveillance, surveillance of people or other moving objects, ambient intelligence applications

The workshop topics will include, but are not limited to, the following:

• Video Surveillance and sensor network architecture
  • Multi-Camera, multi-sensor calibration
  • Camera and/or sensor fusion
  • Cameras and/or sensors coordination and synchronization
  • New sensors, new cameras architectures
  • Self-configurable and Scalable Network Architectures
• Video analysis
  • Deterministic and probabilistic tracking
  • Motion Detection and Tracking
  • Object Detection and Recognition in Unconstrained Environments
  • People behaviour control
  • People posture, gesture, face and interaction analysis
  • Scene recognition
  • Augmented video analysis (with audio and other sensors)
  • Machine Learning Techniques for Event Mining
  • Spatio-temporal Data Mining
• Sensor (Stream) Multimedia Surveillance Data management
  • Annotation, Indexing and Storage
  • Archival/Retrieval of Sensor Data
  • Query Paradigms and Languages

The one and a half day workshop will open with a special session consisting of three to four papers that will introduce and overview the research area. Oral Presentations will then be organized together with possible demo/poster sessions. The accepted papers will be available electronically from the workshop website, and also as printed workshop notes (published by ACM Multimedia).

VSSN 2005 Open Source Algorithm Competition

Organizer: Rainer Lienhart, University of Augsburg, Germany
Committee: Rita Cucchiara, Edward Chang, Eva Hörster [to be finalized]

Web site: www.multimedia-computing.org/VSSN05_OSAC/  (online staring June, 6th 2005)

Progress in the field of video surveillance is slowed down by the lack of open reference implementations and fair competition (i.e., open and public benchmarks) on common problems in video surveillance. Starting this year we will select each year a well-defined surveillance aspect, on which researcher and engineers can compete by focusing on the core aspect only. The supporting source code and data infrastructure will be provided by the competition organizers.

This year’s algorithm competition is on background / foreground segmentation from fixed cameras.

We invite researcher and engineers in the field to submit their algorithm in C/C++ source against a minimal predefined Application Programming Interface (API) consisting of only 3 functions (createBGStatModel(), updateBGStatModel(), releaseBGStatModel()) and one data structure using the OpenCV framework (http://sourceforge.net/projects/opencvlibrary ). Each submitted algorithm must be accompanied by a 4-page paper describing the algorithm. Training sequences, two source code reference implementations, and the source code of the evaluation procedure using the above defined API are provided starting June, 6th 2005. The submissions (deadline 22. Aug 2005) will be tested against unknown, but similar test cases. All complete and working submissions are given the opportunity to present their algorithm as a poster, while the best performing algorithms will get the change to present orally. Furthermore, researcher will have the opportunity to include their source code into next OpenCV release.