|
Ph.D. Defense Thursday, September 20, 2007 ENG II – Room 216 2:00PM Title: Object Tracking and Network Topology Inference in the Multi-modal Sensor Network Abstract: This dissertation presents the human detection and tracking, network topology inference and human activity analysis in the multi-modal sensor network. First, a hierarchical Bayesian network is proposed for detection and tracking of multiple moving objects by fusing the observed video and audio data. Secondly, for camera network topology inference, this dissertation proposes to integrates face recognition that provides robustness to appearance changes and better models the time-varying traffic patterns in the network. Also, a Monte Carlo Expectation Maximization algorithm is employed to continuously learn the network topology and traffic patterns. Finally, the dissertation explores the use of appearance similarity and travel times between departure and arrival nodes to classify anomalous human activities. |
| < Prev | Next > |
|---|