Event Driven Dynamic Context Model for Group Interaction Analysis

Computer understanding of human actions and interactions is the key research issue in human computing. Meanwhile context has been considered to play an essential role in understanding of human behavior during group interactions. This paper proposes a novel Event Driven Dynamic Context Model underlying the context sensing framework to support online analysis of group interactions in meeting scenarios. Sensing of dynamic context is based on multi-level event detection. A novel probabilistic model named Event Driven Multi-level Dynamic Bayesian Network (EDM-DBN) is presented to model hierarchical context and perform online analysis of multi-level events, which is superior over previous works. Experimental results in our smart meeting room demonstrate the effectiveness of our approach.