Interactive unsupervised classication and visualization for browsing an image collection

Pierrick Bruneau (1, 2) , Fabien Picarougne (2) , Marc Gelgon (1, 2)

1 ATLAS - Complex data management in distributed systems
2 LINA - Laboratoire d'Informatique de Nantes Atlantique

In this paper, we propose an approach for interactive navigation in an image collection. As structured groups are more appealing to users than flat image collections, we propose an image clustering algorithm that can handle time-varying collections. A 3D graph-based visualization technique reflects the classification state. While this visualization is itself interactive, we show hos user feedback may assist the classification, thus enabling a user to improve it.

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]

Pierrick Bruneau, Fabien Picarougne, Marc Gelgon. Interactive unsupervised classication and visualization for browsing an image collection. Pattern Recognition, 2010, 43 (2), pp.485-493. ⟨10.1016/j.patcog.2009.03.024⟩. ⟨inria-00396789⟩