Semantic kernel learning for interactive image retrieval

Philippe-Henri Gosselin (1) , Matthieu Cord (2)

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

Content-based image retrieval systems still have difficulties to bridge the semantic gap between the low-level representation of images and the high level concepts the user is looking for. Rele- vance feedback methods deal with this problem using labels pro- vided by users, but only during the current retrieval session. In this paper, we introduce a semantic learning method to manage user la- bels in CBIR applications. Our approach uses a kernel matrix to represent semantic information in a statistical learning framework. The kernel matrix is updated according to labels provided by users after retrieval sessions. Experiments have been carried out on a large generalist database in order to validate our approach.

Domaines: Machine Learning [stat.ML]

Philippe-Henri Gosselin, Matthieu Cord. Semantic kernel learning for interactive image retrieval. IEEE International Conference on Image Processing, Sep 2005, Italy. pp.1. ⟨hal-00520310⟩