Skip to main navigation Skip to search Skip to main content

A deep CNN with focused attention objective for integrated object recognition and localization

  • Xi'an Jiaotong University

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

1 Scopus citations

Abstract

We propose a novel deep convolutional neural network (CNN) architecture able to perform the integrated object recognition and localization tasks. We propose the Focused Attention (FA) objective that aims to optimize the network to learn features only from objects of interest while suppress those features from the background. As a result, the features extracted by the learned models can be used to accurately predict both the object category and the bounding box of the recognized object in the input image. Experimental results show that the proposed CNN architecture trained with the FA objective achieves better performances than original AlexNet in both the object localization and recognition tasks.

Original languageEnglish
Title of host publicationAdvances in Multimedia Information Processing – 17th Pacific-Rim Conference on Multimedia, PCM 2016, Proceedings
EditorsEnqing Chen, Yun Tie, Yihong Gong
PublisherSpringer Verlag
Pages43-53
Number of pages11
ISBN (Print)9783319488950
DOIs
StatePublished - 2016
Event17th Pacific-Rim Conference on Multimedia, PCM 2016 - Xi’an, China
Duration: 15 Sep 201616 Sep 2016

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9917 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference17th Pacific-Rim Conference on Multimedia, PCM 2016
Country/TerritoryChina
CityXi’an
Period15/09/1616/09/16

Keywords

  • AlexNet-FCN
  • CNN
  • FA objective

Fingerprint

Dive into the research topics of 'A deep CNN with focused attention objective for integrated object recognition and localization'. Together they form a unique fingerprint.

Cite this