Efficient Diffusion on Region Manifolds: Recovering Small Objects with Compact CNN Representations

Abstract : Query expansion is a popular method to improve the quality of image retrieval with both conventional and CNN representations. It has been so far limited to global image similarity. This work focuses on diffusion, a mechanism that captures the image manifold in the feature space. The diffusion is carried out on descriptors of overlapping image regions rather than on a global image descriptor like in previous approaches. An efficient off-line stage allows optional reduction in the number of stored regions. In the on-line stage, the proposed handling of unseen queries in the indexing stage removes additional computation to adjust the precomputed data. We perform diffusion through a sparse linear system solver, yielding practical query times well below one second. Experimentally, we observe a significant boost in performance of image retrieval with compact CNN descriptors on standard benchmarks, especially when the query object covers only a small part of the image. Small objects have been a common failure case of CNN-based retrieval.
Type de document :
Communication dans un congrès
2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jul 2017, Honolulu, United States
Liste complète des métadonnées

Littérature citée [64 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01505470
Contributeur : Ahmet Iscen <>
Soumis le : mardi 11 avril 2017 - 13:23:14
Dernière modification le : lundi 16 octobre 2017 - 11:45:43
Document(s) archivé(s) le : mercredi 12 juillet 2017 - 13:12:41

Fichier

Iscen_cvpr17.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01505470, version 1

Collections

Citation

Ahmet Iscen, Giorgos Tolias, Yannis Avrithis, Teddy Furon, Ondřej Chum. Efficient Diffusion on Region Manifolds: Recovering Small Objects with Compact CNN Representations. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jul 2017, Honolulu, United States. 〈hal-01505470〉

Partager

Métriques

Consultations de
la notice

255

Téléchargements du document

127