无监督随机森林.pdf

2013 IEEE International Conference on Computer Vision2013 IEEE International Conference on Computer Vision Video Synopsis by Heterogeneous Multi-Source Correlation 1 2 1 Xiatian Zhu , Chen Change Loy , Shaogang Gong 1Queen Mary, University of London, London E1 4NS, UK 2The Chinese University of Hong Kong, Shatin, N.T., Hong Kong xiatian.zhu@eecs.qmul.ac.uk, ccloy@.hk, sgg@eecs.qmul.ac.uk Abstract Visual Features Non-Visual Data Weather Sensor-based Event Forecast Traffic Data Calendar Generating coherent synopsis for surveillance video stream remains a formidable challenge due to the ambi- guity and uncertainty inherent to visual observations. In contrast to existing video synopsis approaches that rely on visual cues alone, we propose a novel multi-source synopsis Video Synopsis framework capable of correlating visual data and indepen-