Shanghai Jiao Tong University participation in high-level feature extraction, automatic search and surveillance event detectionat TRECVID 2008

In this paper, we describe our participation for high-level feature extraction, automatic search and surveillance event detection at TRECVID 2008 evaluation. In high-level feature extraction, we use selective attention model to extract visual salient feature which highlights the most visual attractive information of an image. Besides this, we extract 7 low-level features for various modalities as a baseline and use linear weighted fusion of multi-modalities . Results show that simple linear weighted fusion works pretty well. In addition, ASR is useful to improve the performance . We submitted the following six runs: a A SJTU 1: Max of all runs based on di erent methods, and Re-rank based on ASR