2014 Poster Sessions : Structural Recognition of Human Activities

Student Name : Tian Lan
Advisor : Silvio Savarese
Research Areas: Artificial Intelligence
Abstract:
Human activity recognition is one of the core problems in computer vision. Recent advances in feature representations and structured learning led to significant progress in this field. With the increasing interest of understanding human activities “in the wild”, new challenges arise due to the complex temporal, spatial, and social structures of real world human activities.

We present a structural approach for understanding human activities in realistic multi-person scenes. Our model describes human behaviour at multiple levels of detail, ranging from low-level actions, groups of people involved in interactions, to high-level events. The hierarchical model includes these varied representations, and various forms of interactions between people present in a scene. We demonstrate our methods in real world human activity applications, including fall detection in nursing home surveillance videos, and automatic annotation of ESPN field hockey broadcast footage.

Bio:
Tian Lan is a post-doctoral researcher in the Computer Science Department at Stanford University. He received the Ph.D. degree in Computer Science from Simon Fraser University (Canada) in 2013. He received his B.Eng from Huazhong University of Science and Technology (China) in 2008. He was a research intern at Disney Research Pittsburgh, in summer 2011 and fall 2012, respectively. His research interests are in computer vision and machine learning, with a focus on visual recognition. He has developed rich structured representations and learning algorithms for semantic understanding of visual scenes from static images as well as video sequences.