Keynotes

Similarity Search in High Dimensional Spaces: Application to Multimedia

Vincent Oria, New Jersey Institute of Technology, USA

Human action recognition: The challenges and recent progress

Ivan Laptev, INRIA Paris - Rocquencourt, France

Multimedia Analytics: Exploration of Large Multimedia Collections

Marcel Worring, Intelligent Systems Lab Amsterdam, University of Amsterdam, The Netherlands

Similarity Search in High Dimensional Spaces: Application to Multimedia

Vincent Oria, New Jersey Institute of Technology, USA

Several novel applications such as recommender systems, information retrieval and multimedia databases operate on high dimensional data, and are designed to process similarity queries that produce ranked lists of objects as their results. Content-based retrieval in Multimedia relies heavily on high dimensional vectors. A typical color histogram is represented with at least 64 dimensions, the popular SIFT descriptor uses 128 dimensions, the ALOI dataset is represented using 641 dimensions and a face is represented with about 2000 dimensions. It is known that data in high dimensional spaces suffer from the so-called curse of dimensionality. However, for those cases where the intrinsic dimension of the data is relatively low, efficient solutions for similarity search are often possible. In this talk, we will survey access methods for similarity search in high-dimensional spaces, and discuss their applications to multimedia.

Short Bio of Vincent Oria

From 1994 to 1996, Vincent Oria worked as a researcher at ENST, Paris; from 1996 to 1999, he was a post-doctoral fellow at the University of Alberta, Edmonton, Canada. From January 2000 to May 2006, Vincent Oria was an assistant professor of computer science at the New Jersey Institute of Technology and since June 2006 he is an Associate Professor in the same department. He has held visiting professor positions at various institutions including National Institute of Informatics (Tokyo, Japan), ENST (Paris, France), Université de Paris-IX Dauphine (Paris, France), INRIA (Roquencourt, France), CNAM (Paris, France), Chinese University of Hong Kong (Hong Kong China) and the Université de Bourgogne (Dijon, France). He is an associate editor for the journals Multimedia Tools and Application (MTAP), the International Journal of Multimedia Data Engineering and Management (IJMDEM) and International Journal of Multimedia Information Retrieval. He has served on a number of multimedia and database conference program committees including ACM Multimedia (MM), ACM World Wide Web (WWW) and IEEE International Conference on Data Engineering (ICDE).

Human action recognition: The challenges and recent progress

Ivan Laptev, INRIA Paris - Rocquencourt, France

Automatic recognition of human actions is a quickly growing research area urged by demands from emerging industries in video indexing, automatic video surveillance, human-computer interaction and other fields. Most applications require action recognition to operate reliably in diverse and realistic video settings. While significant progress towards this goal has been achieved during the last decade, the problem remains highly challenging. 
 This talk will give an overview of the state of the art in action recognition. I will first discuss the problem definition and will argue that action recognition must be addressed within a joint framework together with object recognition and scene understanding. I will then focus on the recent successful methods enabling action recognition in realistic video settings. In particular, I will discuss the new statistical and structural models of actions and the learning issues associated with the training of such models from real data. I will conclude by discussing open questions and promising future research directions.

Short Bio of Ivan Laptev

Ivan Laptev is a full-time researcher in the WILLOW team at INRIA Paris and l'Ecole Normale Superieure (ENS). He received his PhD in Computer Science from the Royal Institute of Technology (KTH) in 2004 and his Master of Science degree from the same institute in 1997. He was a research assistant at the Technical University of Munich (TUM) during 1997-1999 and he has joined INRIA in 2004. Ivan's main research interests concern visual understanding of dynamical scenes including recognition of human actions, scenes and object categories. Ivan has published over 40 papers at international conferences and journals on computer vision, he serves as an associate editor of International Journal of Computer Vision and Image and Vision Computing Journal, he was an area chair for CVPR 2010, ICCV 2011, ECCV 2012 and CVPR 2013.

Multimedia Analytics: Exploration of Large Multimedia Collections 

Marcel Worring, Intelligent Systems Lab Amsterdam, University of Amsterdam, The Netherlands

Multimedia collections, in particular images and videos, are a new and promising source of insight and knowledge in various disciplines. Exploring the content of the collections can only be done when the data is properly analyzed and categorized. When done manually this is a tedious and time-consuming task. Progress in automatic indexing of visual material brings us to a point where tools can help to reduce the effort. As the indexing of visual material is difficult and ambiguous, the quality of the indexing, however, has a limit. As a consequence, much of the categorization is still fully manual. We develop methods which bring the interacting user and the data together through automatic indexing, machine learning, and advanced visualization methods. In this presentation we show the underlying principles of these techniques and how they aid the user in the exploration and categorization processes.

Short Bio of Marcel Worring

Dr. Marcel Worring is Associate Professor of Computer Science at the University of Amsterdam, The Netherlands. His main research interests are in multimedia analytics, bringing together image and video indexing, interactive retrieval, and advanced browsing and visualization. He has published over 150 scientific papers in refereed journals and conferences and was among others co-project leader of the EU Vidivideo project and of the EU Investigator's Dashboard project. He was associate editor of the IEEE Transactions on Multimedia and currently is of the Pattern Analysis and Applications journal. He is guest editor of the upcoming special issue on Learning semantics from Multimedia Web Resources (IEEE Transactions on Multimedia 2012) and of Multimedia in Forensics, Security and Intelligence (IEEE Multimedia 2012). He was the chair of the IAPR TC12 on Multimedia and Visual Information Systems. He was general chair of the Conference on Image and Video Retrieval (CIVR2007), co-initiator and co-organizer of the VideOlympics and co-organizer of the Conference on Multimedia & Expo (ICME, 2005). He is program chair for the ICMR 2013 and the ACM Multimedia 2013.