Workshop Scope

For decades, the computer vision community has been working on content-based multimedia retrieval. Researchers from that community aim at defining better content-based descriptors and extracting them from images. The descriptors obtained are often represented as points in multi-dimensional spaces and some metrics are used during similarity retrieval. Their focus is on increasing the recognition power of their schemes and they usually evaluate their strength using data sets that fit in main memory because they try to avoid the secondary storage management burden.

Facilitating the management of very large amounts of data and removing this disk burden has long been a strong motivation for the database community. This is particularly crucial for multimedia databases whose sizes grow very fast. As such, researchers in databases have proposed many smart multidimensional indexing schemes with some elegant algorithms to compute Nearest-Neighbor and Top-N queries.

Yet, it is surprising to see that only few works in the computer vision community have adopted any of these indexing schemes. A common reason evoked is that the description schemes that database researchers use are way too simplistic. Therefore, it is hard for computer vision researchers to foresee how indexes could behave when used with a modern and powerful description scheme. Additional reasons given include the assumptions on the distribution of data, the ability to only retrieve the single nearest neighbor of query points, and the use of approximate search schemes that give little clue as to the quality of the returned results.

The goal of this workshop is to bridge this gap between the two communities. The idea is to provide database researchers with a snapshot of what computer vision people are dealing with and vice-versa, with the aim of defining some research directions that can benefit both communities. There is great expertise on both sides, and this workshop is aimed at sharing it by means of tutorials and presentations. In addition, we will provide a panel for exchanging ideas with professional image users and/or providers.

Topics of Interest

The main guideline for submitting papers was that they must be at the intersection of the domains mentioned above. We solicited high quality, original papers, which addressed a range of issues in multimedia information retrieval including, but not limited to:

• Content-based indexing, search, and retrieval of multimedia data
• Approximate search techniques and the quality of answers
• High-dimensional clustering for indexing
• Query languages and query processing for multimedia retrieval
• Joint exploitation of multiple media
• Automated semantic content analysis and annotation
• Metadata for multimedia retrieval
• Multi-modal human-computer interaction
• User perspectives
• User modeling
• Color models and image descriptors for very large image sets
• Metrics for data comparison
• Description of compressed documents
• Multimedia data clustering and mining
• Multimedia data modeling and visualization
• Analysis of multimedia data in high-dimensional spaces: statistics, noise, distribution, PCA, skew, ...
• Scalability issues
• Tools, benchmarks, evaluation protocols and standards

Workshop Co-Chairs

• Laurent Amsaleg, IRISA-CNRS, France
• Björn Þór Jónsson, Reykjavík University, Iceland
• Vincent Oria, New Jersey Institute of Technology, USA

Program Committee

• Sibel Adali, Rensselaer Polytechnic Institute, USA
• Marie-Aude Aufaure, Supélec, France
• Ilaria Bartolini, University of Bologna, Italy
• Sid-Ahmed Berrani, IRISA-CNRS, France
• Catherine Berrut, CLIPS-IMAG, France
• Susanne Boll, University of Oldenburg, Germany
• Nozha Boujemaa, INRIA Rocquencourt, France
• Edward Chang, University of California, Santa Barbara, USA
• Chabane Djeraba, LIFL, France
• Valerie Gouet, CNAM Paris - INRIA Rocquencourt, France
• Patrick Gros, IRISA-CNRS, France
• Jarek Gryz, York University, Canada
• Amarnath Gupta, San Diego Supercomputer Center, USA
• Alexander Hinneburg, Martin-Luther-University Halle/Wittenberg, Germany
• Silvia Hollfelder, Fraunhofer IPSI, Germany
• Norio Katayama, National Institute of Informatics, Japan
• Brigitte Kerherve, Université du Québec à Montréal, Canada
• Flip Korn, AT&T Labs-Research, USA
• Francois Le Clerc, Thomson Multimédia R&D, France
• Michael Lew, Leiden University, The Netherlands
• Maude Manouvrier, Paris IX Dauphine University,
• France José Martinez, Ecole Polytechnique de Nantes, France
• Noureddine Mouaddib, Ecole Polytechnique de Nantes, France
• Marco Patella, University of Bologna, Italy
• Eric Pauwels, CWI, Netherlands
• Simone Santini, University of California, San Diego, USA
• Shin'ichi Satoh, National Institute of Informatics, Japan
• Nicu Sebe, University of Amsterdam, The Netherlands
• Cyrus Shahabi, University of Southern California, USA
• James Wang, Penn State University, USA
• Osmar Zaiane, University of Alberta, Canada