CSE Seminar Series

UNIVERSITY of LOUISVILLE

COMPUTER SCIENCE AND ENGINEERING

in cooperation with

IEEE Computer Society ‑ Louisville Chapter

JB Speed Building Room 100

Friday, September 18, 2009

Reception:   3–3:30pm Foyer

Time:  3:30 Seminar

Topic: Semantic Video Indexing and Retrieval

Presenter: Aleksey Fadeev

PhD student, Multimedia Research Lab

CECS Dept University of Louisville

Abstract: In this presentation, we provide an introduction and an overview of the state-of-the-art in video indexing and retrieval. First, we present common methods of audio and images content description based on the low-level features introduced in the MPEG-7 standard. Then, we present our new approach to low-level feature representation based on dominant components. This method, called Dominant Descriptors, allows for more intuitive, compact, and accurate content representation. Examples of several Dominant Descriptors for image and audio content will be defined.

In the second part of the presentation, we outline the Context-Dependent Fusion (CDF) of features and classifiers algorithm developed in our Multimedia Lab. We illustrate the application of this fusion method to content based multimedia retrieval. CDF has two main components: context extraction and algorithm fusion. In context extraction, the low-level audio-visual descriptors used by the different classification algorithms are combined and used to partition the descriptors space into groups of similar video shots, or contexts. The algorithm fusion component identifies a subset of classification algorithms (local experts) for each context based on their relative performance within the context. Finally, an outline of a modern Content Based Multimedia Retrieval System based on multimodal features and multiple classifiers will be presented and discussed.