Kernel on Bags of Fuzzy Regions for fast object retrieval

Philippe-Henri Gosselin (1) , Matthieu Cord (2) , Sylvie Philipp-Foliguet (3)

1 MIDI - Multimedia Indexation and Data Integration
2 MALIRE - Machine Learning and Information Retrieval
3 ETIS - UMR 8051 - Equipes Traitement de l'Information et Systèmes

We propose in this paper a general kernel framework to deal with database object retrieval embedded in images with het- erogeneous background. We use local features computed on fuzzy regions for image representation summarized in bags, and we propose original kernel functions to deal with sets of features and spatial constraints. Combined with SVMs clas- sification and online learning scheme, the resulting algorithm satisfies the robustness requirements for representation and classification of objects. Experiments on a specific database having objects with heterogeneous backgrounds show the per- formance of our object retrieval technique.

Domaines: Machine Learning [stat.ML]

Philippe-Henri Gosselin, Matthieu Cord, Sylvie Philipp-Foliguet. Kernel on Bags of Fuzzy Regions for fast object retrieval. ICIP 2007 - IEEE International Conference on Image Processing, Sep 2007, San Antonio, Texas, United States. pp.177-180, ⟨10.1109/ICIP.2007.4378920⟩. ⟨hal-00520302⟩