Automatic human activity analysis from videos is a very important area of research in computer vision with numerous applications such as surveillance, security, human machine interfaces, sports training, elderly care and monitoring, building smart home and public environments as well as mining web videos for activity based content. From a research point of view, of particular interest are the development of a) rich representations for human motion across several domains, 2) algorithms for tasks such as action recognition and tracking, and c) computationally efficient strategies for performing human activity analysis in very large data sets. In this talk we will address these challenges and propose features and methods for human activity analysis that are very general and can be applied across several domains. The common thread underlying all the methods is the need to explicitly model the temporal dynamics of human motion.
We will first propose optical-flow based and medial-axis skeletal time series features to represent human motion in a scene. We will then model the temporal evolution of these feature time-series using dynamical systems with stochastic inputs and develop methods for comparing these dynamical systems for the purpose of human activity recognition. We will then address the issue of human activity tracking by proposing action-specific dynamic templates. The tracking problem will be posed as a joint optimization problem over the location of the person in the scene as well as the internal state of the dynamical system driving the particular activity. Finally, we will propose very fast approximate-nearest neighbor based methods on the space of dynamical systems for analyzing human motion and show that we can perform human activity recognition very efficiently albeit at a cost of slightly decreased accuracy. Our experimental analysis will show that using dynamical-systems based generative models for human activity perform very well in the above-mentioned tasks.
Speaker Biography
Rizwan Chaudhry received the BSc. (Honors) degree in 2005 with a double major in Computer Science and Mathematics from the Lahore University of Management Sciences in Lahore, Pakistan. Since 2006, he has been pursuing a Ph.D. in the Department of Computer Science at the Johns Hopkins University, where he has been associated with the Vision, Dynamics and Learning Lab under the guidance of Dr. Rene Vidal. His research interests are in the general areas of computer vision and machine learning, and more specifically, modeling dynamic visual phenomena, human activity recognition and tracking.