报告嘉宾1:林巍峣(上海交通大学) 报告时间:2016年6月15日(星期三)晚20:00(北京时间) 报告题目:Trajectory parsing and clustering 主持人: 苏航(清华大学) 报告摘要:Trajectory parsing and clustering is fundamental in many applications including behavior analysis, scene analysis, and video surveillance. In this talk, we will introduce two of our works on trajecory parsing and clustering. First, we introduce a shrinkage-based framework for unsurpervised trajectory clustering. The proposed framework includes an adaptive multi-kernal-based estimation process together with a speed-regularized optimazation process to estimate the shrunk positions and speeds of trajectories' points, such that the discrimination of the shrunk trajectories can be properly increased. Using this approach, the variations among similar trajectories can be reduced while the boundaries between different cluster are enlarged. Second, we further introduce a tube-and-droplet framework for trajetory paring and clustering. Our approach derives scene-related equipotential lines for points in a motion trajectory and concatenate them to construct a 3D tube for representing the trajectory. Based on this 3D tube, a droplet-based method is further proposed which derives a "water droplet" from the 3D tube and recognizes trajectory activities accordingly. Our proposed 3D tube can effectively embed both motion and scene-related information of a motion trajectory. 参考文献: [1] H. Xu, Y. Zhou, W. Lin, H. Zha, "Unsupervised clustering via adaptive multi-kernel-based shrinkage," Intl. Conf. Computer Vision (ICCV), 2015. [2] Y. Zhou, W. Lin, et al. "Representing and recognizing motion trajectories: a tube and droplet approach," ACM Multimedia (MM), 2014. 报告人简介:林巍峣,分别于2003年和2005年获得上海交通大学学士和硕士学位,并于2010年获得美国华盛顿大学西雅图分校获得博士学位。在美国期间曾在包括Motorola, Real Networks和Thomson Technology在内的多家公司的研究机构担任Research Intern。2010年加入上海交通大学电子信息学院电子工程系,现为副教授。主要研究方向包括计算机视觉、视频监控、图像与视频处理、视频通信与编码等。 林博士现任JVCI、Image Comm、IEEE Access等期刊编委(Associate Editor),并任IEEE VSPC TC、IEEE MSA TC、IEEE MMTC等学术专业委员会委员。在相关领域共发表(含录取)SCI期刊论文30+篇(IEEE Transactions系列期刊17篇),会议论文50余篇(ICCV, CVPR, ECCV,MM等9篇);获授权美国发明专利3项,中国发明专利5项。 |
小黑屋|手机版|Archiver|Vision And Learning SEminar
GMT+8, 2024-11-23 10:30 , Processed in 0.012400 second(s), 15 queries .
Powered by Discuz! X3.4
Copyright © 2001-2020, Tencent Cloud.