3rd Workshop on
Statistical Methods in Multi-Image and Video Processing (SMVP) 2006
In conjunction with the 9th European Conference in Computer Vision (ECCV) Graz, Austria.
- On-line submission of full papers open: 10 January 2006
- Closing of on-line submission of full papers: 6 February 2006, 24:00 C.E.T.
- Notification of acceptance: 23 March 2006
- Deadline for submission of full papers: TBA
- Deadline for Camera-ready copies for CD-ROM proceedings (8 pages+multimedia files): 12 April 2006
- Workshop venue: 12 May 2006
A note to authors of accepted SMVP papers
The ECCV oraganizing committee wishes to publish a DVD with all accepted workshop papers, and we in general support this idea. In order not to exclude the possibility to have the SMVP papers also published in a proceedings series (e.g. LNCS) or a journal, we propose the limit the versions of the paper which you submit to the ECCV workshop DVD to 8 pages. We are aware that this is additional work for you, but we think it is worthwhile.
Rudolf Mester and David Suter (on behalf of the SMVP 06 organization committee)
Registration for the SMVP 2006 workshop is managed by ECCV 2006 organizers. To register for the main conference and/or any of the workshops, go to the following page: http://eccv2006.tugraz.at/registration.html. Please note: you do not need to register for the main ECCV conference if you plan to attend only the SMVP 2006 workshop - register only for the workshop(s) you wish to attend.
The registration fee for the SMVP 2006 workshop is 100 EUR for regular participants and 50 EUR for student participants.
Rudolf Mester, Visual Sensorics and Information Processing Lab (VSI)
Goethe-University, Frankfurt, Robert Mayer Str 2-4, 60054 Frankfurt am Main, Germany
- Lourdes Agapito, Queen Mary, University of London, UK
- Patrick Bouthemy, IRISA / INRIA Rennes, France
- Andrew Davison, Imperial College London, UK
- Wolfgang Förstner, Bonn University, Germany
- Anton van den Hengel, Adelaide University, Australia
- Naoyuki Ichimura, AIST, Japan
- Bogdan Matei, Sarnoff Corporation, Princeton, U.S.A.
- Takayuki Okatani, Tohoku University, Japan
- Christoph Schnörr, Mannheim University, Germany
- Nobutaka Shimada, Ritumeikan University, Japan
- Raghav Subbarao, Rutgers University, U.S.A.
- Toshio Ueshiba, AIST, Japan
To contact the organizing commitee please use the adress: email@example.com
In the recent years a variety of advanced statistical methods became standard tools for processing visual information. Our current understanding of the performance of these techniques when applied to lengthy and/or diverse video sequences is, however, rather limited. There are at least two reasons for this. First, the theoretical performance bounds are most often computed with limiting assumptions, which are not valid in practice. Second, robust vision systems require not only superior estimation and analysis to handle outliers and model selection, but also strategies to adapt to the non-stationary behavior of the input.
This workshop focuses on recent progress on the application of modern statistics to modeling and solving computer vision tasks that use multi-image or video data. For such sequences, the underlying models and parameters of the algorithms have to be often adapted or reinitialized in time. Moreover, on-the-fly efficient model selection or detection of redundancy is often necessary.
Examples are sequences containing sudden or gradual changes in the input data statistics, such as: a walking/running/turning person; ego-motion with sudden varying of velocity; empty/crowded train stations; background modeling during the train/metro arrival; turning on/off (either abruptly or gradually) the light source (s); changing the environment from indoor to outdoor; processing an outdoor sequence of clear sky/rain/snow; vision based driver assistance at night; dynamic occlusion modeling; detection and tracking of time-varying patterns; registration of time-varying patterns (medical perfusion, summer/winter surveillance), etc. Treatment of time varying camera parameters, and the detection and handling of degeneracy in structure from multi-view geometry is also within the scope of the workshop.
The workshop aims at bringing together researchers with various backgrounds interested in designing robust vision algorithms that can effectively deal with time-variant statistics of the input video material. Areas of interest include, but are not limited to
- Robust Statistical Techniques for Video Analysis
- Statistical Evaluation of Vision Algorithms
- Tracking and Motion Analysis
- Stereo and Structure from Motion
- 2D and 3D Scene Analysis
- Fluid Motion Analysis
- Background Modeling
- Gesture Recognition
- Video Segmentation and Indexing
- Dynamic Texture Analysis
- Spatio-Temporal Feature Selection
- Industrial Visual Inspection
- Applications in Medical Image Analysis
- Applications in Meteorological Imagery
- Biometrics and Surveillance
However, main workshop themes such as statistical modeling, statistical performance analysis, and adaptation techniques for time-variant statistics should be emphasized.
This event is a sequel to the 2nd Workshop on Statistical Methods in Video Processing, organized in conjunction with the 8th European Conference in Computer Vision in Prague, Czech Republic.
Format, Paper Submission and Review
Original papers with a statistical focus and related to the above topics are solicited.
The submission is electronic, pdf file. The paper should be in English, no longer than 12 pages in Springer LNCS format (same format as for the ECCV 2006). The review is double blind, please do not identify the author(s) in the submission.
It is planned to publish the proceedings in a renowned lecture note series. Details to follow soon.
Best Student Paper Prize
A best student paper prize will be awarded.