About the GPU Research Center at Duke University
The CUDA Research Group at the Department of Computer Science, Duke University includes faculty members who use NVIDIA GPUs and CUDA, in teaching and/or research. In teaching, GPUs are a parallel architecture model and CUDA is used as a programming model. Together, they provide a ready platform for high-performance parallel > computation. Through lectures, seminars, workshops and projects, students get exposed to parallel computation principles, learn how to think parallel in algorithm design and acquire practical skills in parallel programming and computation. In addition to prepare students for their professional placement and development, the knowledge and skills of parallel computation save a direct impact on multiple on-going research projects. Such projects include a) deghosting for stitching gigapixel images, b) long-range video motion analysis, c) automatic inference of the identities of people or vehicles that are tracked by surveillance systems covering wide areas, d) hand tracking to recover 3d pose, e) efficient and robust proximity extraction in high dimensional data spaces, f) highly efficient convolutions and correlations.
About the PIs
The Group members are Nikos Pitsianis, adjunct professor, with PhD in Computer Science from Cornell University and research interests in algorithms and architectures for parallel processing; Xiaobai Sun, professor, with PhD in Computer Science from University of Maryland, College Park and research interests in numerical computation, scientific simulation and image processing; Carlo Tomasi, with PhD in computer science from Carnegie Mellon University and research interests in computer vision, medical imaging, and applied mathematics; Pankaj Agarwal, professor, with PhD in Computer Science from Courant Institute of Mathematical Sciences and research interests in computational and combinatorial geometry, massive data processing, geographic information systems, ecological modeling, computational molecular biology and robotics.