Normalized classifier fusion for semantic visual concept detection

In this paper, we describe our classifier fusion framework for the visual concept detections of NIST TREC-2002 video retrieval benchmark. A normalized ensemble fusion is explored to improve overall performance by incorporating normalization of confidence scores, aggregation via combiner function, and an optimize selection. The normalized classifier fusion shows significant detection improvements for our visual concepts.

[1]  John R. Smith,et al.  VideoAL: a novel end-to-end MPEG-7 video automatic labeling system , 2003, Proceedings 2003 International Conference on Image Processing (Cat. No.03CH37429).

[2]  Timo Ojala,et al.  TREC 2002 Video Track Experiments at MediaTeam Oulu and VTT , 2002, TREC.

[3]  John R. Smith,et al.  A framework for moderate vocabulary semantic visual concept detection , 2003, 2003 International Conference on Multimedia and Expo. ICME '03. Proceedings (Cat. No.03TH8698).

[4]  Haim H. Permuter,et al.  IBM Research TREC 2002 Video Retrieval System , 2002, TREC.

[5]  John R. Smith,et al.  Modeling semantic concepts to support query by keywords in video , 2002, Proceedings. International Conference on Image Processing.

[6]  Junyu Niu,et al.  FDU at TREC 2002: Filtering, Q&A, Web and Video Tasks , 2002, TREC.

[7]  Paul Over,et al.  The TREC-2002 Video Track Report , 2002, TREC.

[8]  Lie Lu,et al.  MSR-Asia at TREC-11 Video Track , 2002 .