论文信息 - CMC-FZU at TRECVID 2010: Semantic Indexing

CMC-FZU at TRECVID 2010: Semantic Indexing

This paper summarizes our approaches to the semantic indexing task in TRECVID 2010. Several fusion strategies are employed in our detection system, including a low-level feature concatenation, text retrieval fusion, and ontology relation fusion. It can be drawn that the fusion of text retrieval results with a baseline in a proper way can contribute greatly to the system performance. Our approach to ontology relation fusion increases the number of the inferred true shots returned. Two of our runs using fusion schemes have the best infAP for a feature among the 30 features evaluated. We describe our text and ontology relation fusion schemes in detail in this paper. The description of our submission runs for the semantic indexing is as follows:

[1] Georges Quénot,et al. LIG at TRECVID 2009: Hierarchical Fusion for High Level Feature Extraction , 2009 .

[2] Chih-Jen Lin,et al. LIBSVM: A library for support vector machines , 2011, TIST.

[3] G LoweDavid,et al. Distinctive Image Features from Scale-Invariant Keypoints , 2004 .

[4] Francesca Odone,et al. Histogram intersection kernel for image classification , 2003, Proceedings 2003 International Conference on Image Processing (Cat. No.03CH37429).

[5] Chen Sun,et al. THU-IMG at TRECVID 2009 , 2009 .

[6] Chao Liang,et al. IVA-NLPR-IA-CAS TRECVID 2009: High Level Features Extraction , 2009 .

[7] Hervé Glotin,et al. IRIM at TRECVID2009: High Level Feature Extraction , 2009 .

[8] Xiaohua Zhai,et al. PKU-ICST at TRECVID 2009: High Level Feature Extraction and Search , 2009 .

[9] Dong Wang,et al. Video diver: generic video indexing with diverse features , 2007, MIR '07.

[10] Chong-Wah Ngo,et al. Columbia University/VIREO-CityU/IRIT TRECVID2008 High-Level Feature Extraction and Interactive Video Search , 2008, TRECVID.

[11] Bernd Freisleben,et al. University of Marburg at TRECVID 2008: High-Level Feature Extraction , 2008, TRECVID.

[12] Jason Hochreiter,et al. UCF @ TRECVID 2009 : High-Level Feature Extraction , 2009, TRECVID.