GSRC Student Profile:
Research Overview: Parallel frameworks for Image Retrieval
My research is in making parallel programming easier for domain experts, with video applications as guiding examples. I aim to provide parallel frameworks, implementations and libraries to aid this effort. I have worked on parallelizing numerous Computer Vision and Machine learning applications including Support vector machines, image contour detection, Large displacement optical flow and Video point tracking. I have also worked on frameworks for optimizing the use of GPU memory when the total application memory footprint exceeds the physical GPU memory. My current work focuses on new computer vision applications such as video object segmentation made possible through the increased computational power available through parallelism and frameworks that help in mapping the nested concurrency in my applications to the hierarchical parallelism found in parallel hardware.
|
