Sciweavers


Publication

Video Synopsis by Heterogeneous Multi-Source Correlation

10 years 5 months ago
Video Synopsis by Heterogeneous Multi-Source Correlation
Generating coherent synopsis for surveillance video stream remains a formidable challenge due to the ambiguity 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 framework capable of correlating visual data and unconstrained non-visual auxiliary information to better describe and summarise subtle physical events in complex scenes. Specifically, our unsupervised framework is capable of seamlessly uncovering latent correlations among heterogeneous types of data sources, despite the non-trivial heteroscedasticity and dimensionality discrepancy problems. The proposed model is robust to partial or missing non-visual information. We demonstrate the effectiveness of our framework on two crowded public surveillance datasets.
X. Zhu, C. C. Loy, and S. Gong
Added 13 Oct 2013
Updated 13 Oct 2013
Type Conference
Year 2013
Where ICCV
Authors X. Zhu, C. C. Loy, and S. Gong
Comments (0)