Image classification with recurrent attention models

Stanislau Semeniuta; Erhardt Barth

doi:10.1109/SSCI.2016.7850113

Image classification with recurrent attention models

Stanislau Semeniuta, Erhardt Barth

Institute of Neuro- and Bioinformatics

Abstract

In this work we apply a fully differentiable Recurrent Model of Visual Attention to unconstrained real-world images. We propose a deep recurrent attention model and show that it can successfully learn to jointly localize and classify objects. We evaluate our model on multiple digit images generated from MNIST data, Google Street View images, and a fine-grained recognition dataset of 200 bird species, and show that its performance is either comparable or superior to that of alternative models.

Original language	English
Title of host publication	2016 IEEE Symposium Series on Computational Intelligence (SSCI)
Publisher	IEEE
Publication date	09.02.2017
Article number	7850113
ISBN (Print)	978-1-5090-4241-8
ISBN (Electronic)	978-1-5090-4240-1
DOIs	https://doi.org/10.1109/SSCI.2016.7850113
Publication status	Published - 09.02.2017
Event	2016 IEEE Symposium Series on Computational Intelligence - Athens, Greece Duration: 06.12.2016 → 09.12.2016 Conference number: 126460

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/SSCI.2016.7850113

Cite this

@inproceedings{b31c9a659d61421393fd0cde1dbc116b,

title = "Image classification with recurrent attention models",

abstract = "In this work we apply a fully differentiable Recurrent Model of Visual Attention to unconstrained real-world images. We propose a deep recurrent attention model and show that it can successfully learn to jointly localize and classify objects. We evaluate our model on multiple digit images generated from MNIST data, Google Street View images, and a fine-grained recognition dataset of 200 bird species, and show that its performance is either comparable or superior to that of alternative models.",

author = "Stanislau Semeniuta and Erhardt Barth",

note = "Publisher Copyright: {\textcopyright} 2016 IEEE. Copyright: Copyright 2017 Elsevier B.V., All rights reserved.; 2016 IEEE Symposium Series on Computational Intelligence, SSCI 2016 ; Conference date: 06-12-2016 Through 09-12-2016",

year = "2017",

month = feb,

day = "9",

doi = "10.1109/SSCI.2016.7850113",

language = "English",

isbn = "978-1-5090-4241-8",

booktitle = "2016 IEEE Symposium Series on Computational Intelligence (SSCI)",

publisher = "IEEE",

}

TY - GEN

T1 - Image classification with recurrent attention models

AU - Semeniuta, Stanislau

AU - Barth, Erhardt

N1 - Conference code: 126460

PY - 2017/2/9

Y1 - 2017/2/9

N2 - In this work we apply a fully differentiable Recurrent Model of Visual Attention to unconstrained real-world images. We propose a deep recurrent attention model and show that it can successfully learn to jointly localize and classify objects. We evaluate our model on multiple digit images generated from MNIST data, Google Street View images, and a fine-grained recognition dataset of 200 bird species, and show that its performance is either comparable or superior to that of alternative models.

AB - In this work we apply a fully differentiable Recurrent Model of Visual Attention to unconstrained real-world images. We propose a deep recurrent attention model and show that it can successfully learn to jointly localize and classify objects. We evaluate our model on multiple digit images generated from MNIST data, Google Street View images, and a fine-grained recognition dataset of 200 bird species, and show that its performance is either comparable or superior to that of alternative models.

UR - https://www.scopus.com/record/display.uri?eid=2-s2.0-85016005598&origin=inward&txGid=fd7a415b5b82b3e67788fd6ffe557c01

U2 - 10.1109/SSCI.2016.7850113

DO - 10.1109/SSCI.2016.7850113

M3 - Conference contribution

SN - 978-1-5090-4241-8

BT - 2016 IEEE Symposium Series on Computational Intelligence (SSCI)

PB - IEEE

T2 - 2016 IEEE Symposium Series on Computational Intelligence

Y2 - 6 December 2016 through 9 December 2016

ER -

Image classification with recurrent attention models

Abstract

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this