File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Conference Paper: Aggregated Deep Activation Clusters for Particular Object Retrieval

TitleAggregated Deep Activation Clusters for Particular Object Retrieval
Authors
Issue Date2017
PublisherACM.
Citation
Thematic Workshops '17: Proceedings of the on Thematic Workshops of ACM Multimedia (MM) 2017, Mountain View, California, USA, 23-27 October 2017, p. 44-51 How to Cite?
AbstractThis paper introduces a clustering based deep feature for particular object retrieval. Many object retrieval algorithms focus on aggregating local features into compact image representations. Recently proposed algorithms, such as R-MAC and its variants, aggregate maximum activations of convolutions from rectangular regions of multiple scales and have achieved state-of-the-art performance. Such rectangular regions, however, cannot fit the 'non-rectangular' shape of an arbitrary object well, and therefore cover much clutter in the background. This paper targets at mitigating this problem by proposing a deep feature based on clustering the activations of convolutions and aggregating the maximum activations from such clusters. Compared with the square regions used in R-MAC, the clusters thus obtained can better fit the arbitrary shapes and sizes of the objects of interest. By not taking spatial location into account, it is possible to have a single cluster covering multiple disconnected regions that correspond to repeated but isolated visual patterns. This helps to avoid over-weighting such patterns in the aggregated feature. Experiments are carried out on the challenging Oxford5k and Paris6k datasets, and results show that our clustering based deep feature outperforms the R-MAC feature.
Persistent Identifierhttp://hdl.handle.net/10722/246606
ISBN

 

DC FieldValueLanguage
dc.contributor.authorChen, Z-
dc.contributor.authorKuang, Z-
dc.contributor.authorWong, KKY-
dc.contributor.authorZhang, W-
dc.date.accessioned2017-09-18T02:31:27Z-
dc.date.available2017-09-18T02:31:27Z-
dc.date.issued2017-
dc.identifier.citationThematic Workshops '17: Proceedings of the on Thematic Workshops of ACM Multimedia (MM) 2017, Mountain View, California, USA, 23-27 October 2017, p. 44-51-
dc.identifier.isbn978-1-4503-5416-5-
dc.identifier.urihttp://hdl.handle.net/10722/246606-
dc.description.abstractThis paper introduces a clustering based deep feature for particular object retrieval. Many object retrieval algorithms focus on aggregating local features into compact image representations. Recently proposed algorithms, such as R-MAC and its variants, aggregate maximum activations of convolutions from rectangular regions of multiple scales and have achieved state-of-the-art performance. Such rectangular regions, however, cannot fit the 'non-rectangular' shape of an arbitrary object well, and therefore cover much clutter in the background. This paper targets at mitigating this problem by proposing a deep feature based on clustering the activations of convolutions and aggregating the maximum activations from such clusters. Compared with the square regions used in R-MAC, the clusters thus obtained can better fit the arbitrary shapes and sizes of the objects of interest. By not taking spatial location into account, it is possible to have a single cluster covering multiple disconnected regions that correspond to repeated but isolated visual patterns. This helps to avoid over-weighting such patterns in the aggregated feature. Experiments are carried out on the challenging Oxford5k and Paris6k datasets, and results show that our clustering based deep feature outperforms the R-MAC feature.-
dc.languageeng-
dc.publisherACM.-
dc.relation.ispartofACM International Conference on Multimedia (MM) Thematic Workshops-
dc.titleAggregated Deep Activation Clusters for Particular Object Retrieval-
dc.typeConference_Paper-
dc.identifier.emailWong, KKY: kykwong@cs.hku.hk-
dc.identifier.authorityWong, KKY=rp01393-
dc.identifier.doi10.1145/3126686.3126696-
dc.identifier.hkuros276755-
dc.identifier.spage44-
dc.identifier.epage51-
dc.publisher.placeNew York, NY-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats