 Perception is the process of using sensory information to gain an understanding about the external world. "That is, we are constantly using our human senses - vision, audition, touch, smell - to derive meaning from the environment around us," says Visiting Assistant Professor of Computer Science Doug Turnbull. "This may include seeing a deer in the woods, identifying your mother's voice on the phone, noting that a wool sweater is scratchy, or smelling a warm apple pie." Each of these basic activities involves a number of basic tasks. "For example, when we spot a deer in the words we must segment the object (deer) from the background (woods)," he says. "We then need to identify the objects as deer. We might cluster one deer with other deer nearby, then track the deer as they move across the road. Based on our perception of the deer, we then may decide to slam on the breaks or honk our car horn." In Doug's seminar on Computer Perception, he and his students explored various computer systems that are able to derive meaning from sensors (e.g., cameras & microphones). "We have primarily focused on computer vision tasks - such as segmenting and identifying faces in images, tracking moving objects in videos, and rendering 3-D geometries based on multiple 2-D images - and computer audition tasks - including converting human speech to text, classifying music by genre, and locating the source of a sound," he says. "What follows is what we found." |
Visiting Assistant Professor of Computer Science Douglas Turnbull joined Swarthmore's faculty this fall. His main research interests include computer audition, machine learning, music information retrieval, and human computation. Doug recently earned a Ph.D. in computer science from UC-San Diego, where he was an NSF IGERT Fellow. While at UCSD, he co-founded the interdisciplinary Computer Audition Laboratory. Write to him at dturnbu1@swarthmore.edu
|