Date: February 19, 2009 (Thursday)
Time: 4:45pm (refreshment starts at 4:30pm)
Place: 202 ECEC, NJIT
About the Presenter:
Alexander C. Loui obtained his B.A.Sc. (Honors), M.A.Sc, and Ph.D. all in Electrical and Computer Engineering from the University of Toronto, Canada. In 1990, he joined Bellcore as a Member of Technical Staff working on audiovisual compression and VOD technologies. He joined Kodak Research Labs in 1996. Since then he has led and contributed to pioneering research on digital image management including event detection, auto-albuming algorithms, and multimedia composition systems. His research interests span the areas of semantic content understanding, multimedia indexing and retrieval, intelligent systems, and video communications. He has been an associate editor of the IEEE Transactions on Multimedia, IEEE Transactions on Circuits and Systems, and SPIE Journal of Electronic Imaging. He was Chair (2005) of the Rochester Chapter of IEEE Signal Processing Society. He has been the Vice Chair (2009) and Treasurer (2007-08) of the IEEE Rochester Section. He is a Kodak Distinguished Inventor (with over 50 granted and pending patents), and a Fellow of IEEE “for contributions to digital image content management systems.”
About the Talk:
The increase popularity of smart media devices and social networking has led to an explosion in the amount of digital media content being created. This has resulted in large personal and public multimedia databases in which it has become increasingly difficult to retrieve specific content and browse the large collections. Current content management systems only provide simple browsing and navigation capability based on manual annotations, which severely limits the search and other advanced functionality. Our research is focused on semantic content understanding and analysis to enable easy browsing, searching, composing, and sharing of content and personal memories. In this talk, I will describe some recent work on semantic event detection and image value indexing.
We propose a novel semantic event detection approach by considering an event-level Bag-of-Features (BOF) representation to model typical consumer events. Based on this BOF representation, semantic events are detected in a concept space instead of the original low-level visual feature space. There are two advantages to our approach: we can avoid the sensitivity problem by decreasing the influence of difficult or erroneous images or videos in measuring eventlevel similarity; also we can utilize the power of higher-level concept scores in describing semantic events. The ability to automatically assess image characteristics is another important function for content management, building photo albums, and retrieval of specific visual content. We proposes a novel approach to assess and rate images based on multidimensional characteristics including image quality, social relationships, aesthetic quality, important events, and usage. This new approach provides additional flexibility for end user applications that utilize different aspects of image characteristics. Specifically, we describe a method for assessing image quality based upon technical characteristics of the image, and for predicting the significance of an image based upon the people portrayed in the image.
Note: All MS thesis and PhD dissertation (proposal) defense are counted towards ECE791.