Illuminant estimation from shadows typically relies on accurate segmentation of the shadows and knowledge of exact 3D geometry, while shadow estimation is difficult in the presen...
This work introduces two complementary techniques to improve the packet loss resilience of scalable video coding systems. First, a “switch per-pixel” error concealment (SPEC) ...
There is a lack of an integrated technology that will increase effective usage of the vast and heterogeneous multi-lingual and multimedia digital content. The need is being express...
The explosion of Web opinion data has made essential the need for automatic tools to analyze and understand people’s sentiments toward different topics. In most sentiment analy...
We combine local texture features (PCA-SIFT), global features (shape context), and spatial features within a single multi-layer AdaBoost model of object class recognition. The fir...
Wei Zhang 0002, Bing Yu, Gregory J. Zelinsky, Dimi...