When someone asks you “how was your day?” we usually indicate only the important events of the day to serve an understandable summary. We don’t talk about breathing, every step we take, or anything too detailed that we judge as insignificant to our day. Although this ability is inherent for humans, extensive coding is needed for computers to perform the same task.
Assistant Professor Yong Jae Lee, one of the newest members of UCD’s Computer Science Department, has been developing coding that can accurately and concisely summarize important information from a video. He graduated from the University of Illinois Urbana-Champaign with a B.S. in Electrical and Computer Engineering, and got his Master’s degree in Electrical and Computer Engineering and Ph.D at the University of Texas, Austin. He has published several papers focusing on computer vision, most recently the 2014 study, “Predicting Important Objects for Egocentric Video Summarization.”
The following is adapted from an email interview with Assistant Professor Lee.
The Centennial: Many of your studies focus on computer vision. Can you elaborate on the field of study?
Assistant Professor Yong Jae Lee: Computer vision is the study of building machines that can “see” the way we humans do. My research focuses on the design of algorithms and image representations that allow a machine to accurately recognize objects, their properties and the activities of people appearing in an image or video with minimal human guidance.
TC: In your study “Predicting Important Objects for Egocentric Video Summarization” what were your main findings? What are the implications?
YJL: In that work, we developed an algorithm that automatically creates a short visual summary of a very long video taken by a wearable camera, like Google Glass. Our algorithm predicts the important people and objects that the camera wearer interacted with, and uses those predictions to select the keyframes that go into the final summary. With our algorithm, we can go from a video that is several hours long to a short keyframe summary that depicts the key happenings of the camera wearer’s day, and can be viewed in a matter of seconds.
TC: In what ways will this research be applied in society?
YJL: There have been studies that show that people with memory problems can better recall key events that happened in their day when viewing a visual diary compared to when reading a written diary. Our algorithm could produce visual diaries to aid such patients. Similarly, our algorithm could be used to summarize videos captured by police officers while on duty or to summarize surveillance videos. Another application could be to summarize videos captured by robots that are exploring new unexplored territories. These would help relieve the burden of someone having to watch these videos from beginning to end, which can sometimes be many hours long.
TC: What should be done to increase interest in computer science?
YJL: To increase interest in any discipline, I feel it’s important to provide accessibility and opportunities to people from a young age. Computer science is no different. This means providing computer classes in elementary, middle and high school.
TC: What important aspects do you believe students should know about computer science?
YJL: The field of computer science is growing rapidly; computer scientists will be in much need, as computers will become ubiquitous in almost all aspects of our everyday lives. If you are a creative person that likes to solve problems and think clearly and logically, then this is a field that you should definitely consider.
TC: Aside from your area of research, what other fields within computer science interest you?
YJL: While my current research interests are in computer vision, machine learning, and computer graphics, I’m generally interested in all areas of Artificial Intelligence. One research area that I would like to explore in the future is robotics. I want to create a robot that can autonomously learn about environment by physically interacting with objects, like grabbing them, rotating them, etc., much like how we humans learn about our world.
For more information, visit Assistant Professor Yong Jae Lee’s website at web.cs.ucdavis.edu/~yjlee.
Tarischka Stamboel is a staff writer for The Centennial Magazine. She can answer any questions at email@example.com