Utilisation de concepts visuels et de la diversité visuelle pour améliorer la recherche d'images

In this article, we study (i) how to automatically extract an d exploit visual concepts and (ii) fast visual diversity. First, in the Visual ConceptDetection Task (VCDT), we look at the mutual exclusion and implication relations between VCD T concepts in order to improve the automatic image annotation by Forest of Fuzzy Decision T rees (FFDTs). Second, in the ImageCLEFphoto task, we use the FFDTs learnt in VCDT task and WordNet to improve image retrieval. Third, we apply a fast visual diversity method ba sed on space clustering to improve the cluster recall score. This study shows that there is a cle ar improvement, in terms of precision or cluster recall at 20, when using the visual concepts expli citly appearing in the query and that space clustering can be efficiently used to improve cluster r