Apprentissage de métrique appliqué à la détection de changement de page Web et aux attributs relatifs

This paper introduces a novel distance metric learning framework. Working with inequality constraints involving quadruplets of images, our approach aims at efficiently modeling similarity for rich or complex semantic label relationships. We study how our metric learning scheme can be exploited in contexts such as detection of important regions in Webpages or recognition with relative attributes. MOTS-CLÉS : apprentissage de métrique, reconnaissance d’image, détection de changement