I do research in computer vision, working on estimating motion in video sequences. Extracted motion information is important because it can be used to solve higher-level vision tasks such as segmenting objects of interest or recognizing events. My work focuses on estimating motion by considering many frames of a video clip at once. Looking at a block of video, rather than just pairs of frames, lets us exploit long-range temporal information to find and reason about occlusion surfaces, which are the truly challenging part of motion estimation. In constructing these systems, we get to test our theories about how the world works and how we make decisions about the world. I'm intrigued by the problems that we accomplish so easily (like recognizing a bird in an image) but that are difficult to express mathematically so that a computer can accomplish them.
I spent the summer working at MIT Lincoln Laboratory in Lexington, MA and got to apply my research on inferring motion to data with interesting properties. I worked with a fantastic team on a project with a clear impact. This is what I'd like to do when I graduate.
The thing I like most about our department is the sense of community we have here. Doing research and getting a Ph.D. can be very challenging at times and it is vital to have a group of people supporting you who are committed to helping you succeed. Many Ph.D. programs admit far more students than will graduate, intending to weed them out. Duke only admits students that they expect will succeed, and our administration focuses on supporting our research, not putting roadblocks between us and success.
Return to CS@DUKE WOMEN page