I am a 5th year PhD student in the department of Computing Science in University of Alberta. I have passed my candidacy examination last year. I am currently in the final stage of my research and planning to defend by next semester. In my PhD thesis, I have developed a novel people counting algorithm for computing unique people count from monocular videos. The algorithm has the capability of handling severe occlusion in addition to computing unique people count with exorbitant accuracy. Also it is online in nature, and does not accumulate error over time. I have performed extensive experiments with the proposed algorithm on four standard datasets - the UCSD dataset (Chan et al., 2008), which consist of a full one hour video of 25,656 frames, the FUDAN dataset (Tan et al., 2011) consisting of 1500 frames, the LHI dataset (Cong et al., 2009) which has 12 videos captured at different camera angles (90 degree, 65 degree and 40 degree) and of duration between 5 minutes and 15 minutes, and the PETS 2009 dataset (Krahnstoever et al., 2008) consisting of multiple camera views, targeted at the evaluation of various surveillance applications. The algorithm has produced more than 95% accuracy for most of these videos.
展开▼
