Skip to Main Content
This paper proposes a method for event detection based on hierarchical event fusion. Four high-level events in intelligent meeting scenarios, namely, ldquomonologuerdquo,ldquopresentationrdquo, ldquodiscussionrdquo, and ldquobreakrdquo, are analyzed. To characterize these four events by hierarchical event fusion and inference, four kinds of group events are considered. Group events are analyzed based on three kinds of basic states of individual participants, such as location, standing or sitting, and speaking or silence. Rao-Blackwellized particle filters are applied to make event inference in real time. The experimental results indicate that this approach is effective in detecting high-level event.