Object Localization with Attribute Preference Based on Top-Down Attention

  • Soubarna Banik
  • , Mikko Lauri
  • , Alois Knoll
  • , Simone Frintrop

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We propose a weakly-supervised approach for object localization based on top-down attention which is able to consider both attributes and object classes as attentional cues. This enables to not only search for objects but additionally for objects with specific attributes such as colors or shapes. Our approach consists of two streams: an attribute stream and an object stream. By tracing backward through these two streams and localizing activated neurons in hidden layers, we generate two top-down attention maps, one for attributes and one for objects. Fusing these maps generates a joint attention map, which highlights regions with a specific attribute and object. We show experimentally that our method can localize objects in cluttered images by only specifying their attributes, and that instances of the same class can be discriminated based on their attributes.

Original languageEnglish
Title of host publicationComputer Vision Systems - 13th International Conference, ICVS 2021, Proceedings
EditorsMarkus Vincze, Timothy Patten, Henrik I Christensen, Lazaros Nalpantidis, Ming Liu
PublisherSpringer Science and Business Media Deutschland GmbH
Pages28-40
Number of pages13
ISBN (Print)9783030871550
DOIs
StatePublished - 2021
Event13th International Conference on Computer Vision Systems, ICVS 2021 - Virtual, Online
Duration: 22 Sep 202124 Sep 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12899 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference13th International Conference on Computer Vision Systems, ICVS 2021
CityVirtual, Online
Period22/09/2124/09/21

Keywords

  • Object attribute
  • Object localization
  • Top-down attention

Fingerprint

Dive into the research topics of 'Object Localization with Attribute Preference Based on Top-Down Attention'. Together they form a unique fingerprint.

Cite this