Open tutorial - From Computer Vision to Databases

Reykjavík University Database Lab welcomes you to an open tutorial named "From Computer Vision to Databases" on September 23. and 25.  The presenters are scientists from the IRISA research lab in Rennes, France, who have specialized in image database research for many years, partially in collaboration with the RU Database Lab.

The tutorial will be held in room O-131b at RU's facilities in Ofanleiti 2, as follows:

- Tuesday 23.9, at 15:30 - 19:00
- Thursday 25.9, at 15:30 - 19:00

The tutorial is open to all interested parties, but we kindly ask you to register by sending mail to skraning@ru.is.

A description of the tutorial follows, as well as biographies of the presenters.

Welcome,
Björn Þór Jónsson
Director of the RU Database Lab

From Computer Vision to Databases
- an introduction to image description, indexing and query processing.

The tutorial will introduce and explore the following topics:

Image Descriptions:
- Introduction and motivation
- The basics of image description: color, texture, shape.
- Advanced description: interest points and local descriptors.

Image Indexing and Querying:
- introduction and motivation
- typical tasks: searching in a high-dimensional neighborhood
- the need for indexing
- cells, geometry, filtering
- initial approaches: R-trees, KD-trees
- properties of high-dimensional spaces
- the coolest idea: going approximate
- recent approaches, including LSH, OMEDRANK and NV-Tree

Application to copy detection:
- Visual words and Video Google.
- Descriptor evaluation.

Biographies:

Patrick GROS has been involved in research in the field of image analysis for 18 years. He received his PhD degree from University of Grenoble - France in 1993. After six years in the GRAVIR lab in Grenoble and one year at the Robotics Institute of CMU (Pittsburgh, PA, USA), he moved in 1999 to the IRISA lab of Rennes - France where he worked in the VISTA research group. In 2002, he founded the new TEXMEX group dedicated to multimedia indexing and retrieval. The key idea of the team is to gather specialists of various media and techniques (learning, databases, data analysis) in a unique team in order to cope with problem like multimodality and coupling between document description and indexing techniques. Patrick Gros's research interests concern multimedia indexing  and retrieval in very large collections with applications like copy detection, TV analysis, audiovisual information retrieval.

Laurent Amsaleg received his PhD in June 1995 from the University of Paris 6, France. He worked on relational and object-oriented databases, garbage collection, micro-kernels and single-level stores. He then spent 18 months in the Database group of the University of Maryland (MD, USA), designing flexible database query execution strategies (Query Scrambling). Subsequently, he received a full-time research position at CNRS in France and joined the IRISA lab in Rennes. He works with the TEXMEX group. His research primarily focuses on the performance of content based multimedia retrieval, which include the design of multidimensional indexing techniques. He was co-creator of the CVDB (Computer vision meets Databases) workshop series.