Patch-level spatial layout for classification and weakly supervised localization

Valentina Zadrija 1 Josip Krapac 1 Jakob Verbeek 2 Siniša Šegvić 1
2 LEAR - Learning and recognition in vision
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
Abstract : We propose a discriminative patch-level spatial layout model suitable for training with weak supervision. We start from a block-sparse model of patch appearance based on the normalized Fisher vector representation. The appearance model is responsible for i) selecting a discriminative subset of visual words, and ii) identifying distinctive patches assigned to the selected subset. These patches are further filtered by a sparse spatial model operating on a novel representation of pairwise patch layout. We have evaluated the proposed pipeline in image classification and weakly supervised localization experiments on a public traffic sign dataset. The results show significant advantage of the proposed spatial model over state of the art appearance models.
Type de document :
Communication dans un congrès
Juergen Gall; Peter Gehler; Bastian Leibe. 37th German Conference on Pattern Recognition, GCPR 2015, Oct 2015, Aachen, Germany. Springer, 9358, pp.492-503, Lecture Notes in Computer Science. <http://gcpr2015.rwth-aachen.de/>. <10.1007/978-3-319-24947-6_41>
Liste complète des métadonnées


https://hal.inria.fr/hal-01186677
Contributeur : Thoth Team <>
Soumis le : jeudi 27 août 2015 - 15:12:25
Dernière modification le : jeudi 9 février 2017 - 16:12:03
Document(s) archivé(s) le : samedi 28 novembre 2015 - 10:19:52

Fichier

main68.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

Collections

Citation

Valentina Zadrija, Josip Krapac, Jakob Verbeek, Siniša Šegvić. Patch-level spatial layout for classification and weakly supervised localization. Juergen Gall; Peter Gehler; Bastian Leibe. 37th German Conference on Pattern Recognition, GCPR 2015, Oct 2015, Aachen, Germany. Springer, 9358, pp.492-503, Lecture Notes in Computer Science. <http://gcpr2015.rwth-aachen.de/>. <10.1007/978-3-319-24947-6_41>. <hal-01186677>

Partager

Métriques

Consultations de
la notice

262

Téléchargements du document

254