Friday, March 6, 2015 - 9:00am to 10:00am
KEC 1007

Speaker Information

Michael S. Ryoo
Research Staff
NASA Jet Propulsion Laboratory

Abstract

We are entering the era of big video data where cameras are ubiquitous. In particular, the amount of videos from wearable cameras and robots is explosively rising. These videos, taken from an actor's own viewpoint, are called 'first-person videos' or 'egocentric videos'. Millions of individuals are already recording their lives using wearable cameras and, soon, robots in public places will obtain similar videos capturing their operations and interactions in the world. This talk presents automated methodologies to make sense of all this visual data by detecting important events in videos and generating compact summaries that describe these events. This will not only allow video-based robot learning and recognition, but also enable construction of intelligent wearable systems supporting human tasks such as medical operations, law enforcement, activities of daily living, and human-robot teaming. We discuss features and recognition algorithms necessary for 'activity-level' understanding of such first-person videos, and describe how they make recognition of human-human (and human-robot) interactions possible. Approaches for early recognition of ongoing activities from streaming videos will be described, and the future scenario of multiple wearable/robot cameras and static cameras (e.g., surveillance cameras in smart cities) coexisting will be discussed.

Speaker Bio

Michael S. Ryoo is a research staff at NASA's Jet Propulsion Laboratory. His research interests are in computer vision and robotics, including semantic understanding of video data, first-person vision, and intelligent interaction/collaboration between humans and wearbles/robots. Dr. Ryoo received the B.S. degree in computer science from Korea Advanced Institute of Science and Technology (KAIST) in 2004, and the M.S. and Ph.D. degrees in computer engineering from the University of Texas at Austin in 2006 and 2008, respectively. He has authored a number of pioneering papers on human activity recognition, has been providing tutorials on activity recognition at major computer vision conferences including CVPR 2011, AVSS 2012, and CVPR 2014, and is the corresponding author of the activity recognition survey paper published by ACM Computing Surveys on 2011. He organized the first ICPR contest on human activity recognition (SDHA 2010) and the 3rd workshop on Egocentric Vision at CVPR 2014.