IRIM at TRECVID2009: High Level Feature Extraction

Bertrand Delezoide 1 Hervé Le Borgne 1 Pierre-Alain Moëllic 1 David Gorisse 2 Frédéric Precioso 2 Feng Wang 3 Bernard Merialdo 3 Philippe-Henri Gosselin 4 Lionel Granjon 5 Denis Pellerin 6 Michèle Rombaut 6 Hervé Bredin 7 Lionel Koenig 7 Hélène Lachambre 7 Elie El Khoury 7 Boris Mansencal 8 Yifan Zhou 8 Jenny Benois-Pineau 8 Hervé Jégou 9 Stéphane Ayache 10 Bahjat Safadi 11 Georges Quénot 11 Jonathan Fabrizio 12 Matthieu Cord 2, 12 Hervé Glotin 13 Zhongqiu Zhao 13 Emilie Dumont 13 Bertrand Augereau 14
Abstract : The IRIM group is a consortium of French teams working on Multimedia Indexing and Retrieval. This paper describes our participation to the TRECVID 2009 High Level Features detection task. We evaluated a large number of different descriptors (on TRECVID 2008 data) and tried different fusion strategies, in particular hierarchical fusion and genetic fusion. The best IRIM run has a Mean Inferred Average Precision of 0.1220, which is significantly above TRECVID 2009 HLF detection task median performance. We found that fusion of the classification scores from different classifier types improves the performance and that even with a quite low individual performance, audio descriptors can help.
Document type :
Conference papers
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-00468199
Contributor : Boris Mansencal <>
Submitted on : Tuesday, March 30, 2010 - 12:14:54 PM
Last modification on : Friday, April 12, 2019 - 4:23:00 PM
Document(s) archivé(s) le : Thursday, July 1, 2010 - 8:25:41 PM

File

IRIMatTRECVID2009.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00468199, version 1

Citation

Bertrand Delezoide, Hervé Le Borgne, Pierre-Alain Moëllic, David Gorisse, Frédéric Precioso, et al.. IRIM at TRECVID2009: High Level Feature Extraction. TREC Video Retrieval Evaluation: TRECVID, Nov 2009, Gaithersburg, MD, United States. ⟨hal-00468199⟩

Share

Metrics

Record views

3555

Files downloads

488