Collections for automatic image annotation and photo tag recommendation

Philip J. Mcparlane, Yashar Moshfeghi, Joemon M. Jose

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

4 Citations (Scopus)

Abstract

This paper highlights a number of problems which exist in the evaluation of existing image annotation and tag recommendation methods. Crucially, the collections used by these state-of-the-art methods contain a number of biases which may be exploited or detrimental to their evaluation, resulting in misleading results. In total we highlight seven issues for three popular annotation evaluation collections, i.e. Corel5k, ESP Game and IAPR, as well as three issues with collections used in two state-of-the-art photo tag recommendation methods. The result of this paper is two freely available Flickr image collections designed for the fair evaluation of image annotation and tag recommendation methods called Flickr-AIA and Flickr-PTR respectively. We show through experimentation and demonstration that these collection are ultimately fairer benchmarks than existing collections.
LanguageEnglish
Title of host publicationMultiMedia Modeling
Subtitle of host publication20th Anniversary International Conference, MMM 2014, Dublin, Ireland, January 6-10, 2014, Proceedings, Part I
Place of PublicationNew York, NY
Pages133-145
Number of pages13
DOIs
Publication statusPublished - 20 Dec 2013
EventInternational Conference on Multimedia Modeling - Dublin, Ireland
Duration: 6 Jan 201410 Jan 2014

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume8325

Conference

ConferenceInternational Conference on Multimedia Modeling
CountryIreland
CityDublin
Period6/01/1410/01/14

Fingerprint

Demonstrations

Keywords

  • image annotation
  • tag recommendation methods
  • evaluation
  • social media

Cite this

Mcparlane, P. J., Moshfeghi, Y., & Jose, J. M. (2013). Collections for automatic image annotation and photo tag recommendation. In MultiMedia Modeling: 20th Anniversary International Conference, MMM 2014, Dublin, Ireland, January 6-10, 2014, Proceedings, Part I (pp. 133-145). (Lecture Notes in Computer Science; Vol. 8325). New York, NY. https://doi.org/10.1007/978-3-319-04114-8_12
Mcparlane, Philip J. ; Moshfeghi, Yashar ; Jose, Joemon M. / Collections for automatic image annotation and photo tag recommendation. MultiMedia Modeling: 20th Anniversary International Conference, MMM 2014, Dublin, Ireland, January 6-10, 2014, Proceedings, Part I. New York, NY, 2013. pp. 133-145 (Lecture Notes in Computer Science).
@inproceedings{b63e4471343a4f33a594e7cf07819050,
title = "Collections for automatic image annotation and photo tag recommendation",
abstract = "This paper highlights a number of problems which exist in the evaluation of existing image annotation and tag recommendation methods. Crucially, the collections used by these state-of-the-art methods contain a number of biases which may be exploited or detrimental to their evaluation, resulting in misleading results. In total we highlight seven issues for three popular annotation evaluation collections, i.e. Corel5k, ESP Game and IAPR, as well as three issues with collections used in two state-of-the-art photo tag recommendation methods. The result of this paper is two freely available Flickr image collections designed for the fair evaluation of image annotation and tag recommendation methods called Flickr-AIA and Flickr-PTR respectively. We show through experimentation and demonstration that these collection are ultimately fairer benchmarks than existing collections.",
keywords = "image annotation, tag recommendation methods, evaluation, social media",
author = "Mcparlane, {Philip J.} and Yashar Moshfeghi and Jose, {Joemon M.}",
year = "2013",
month = "12",
day = "20",
doi = "10.1007/978-3-319-04114-8_12",
language = "English",
isbn = "9783319041131",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "133--145",
booktitle = "MultiMedia Modeling",

}

Mcparlane, PJ, Moshfeghi, Y & Jose, JM 2013, Collections for automatic image annotation and photo tag recommendation. in MultiMedia Modeling: 20th Anniversary International Conference, MMM 2014, Dublin, Ireland, January 6-10, 2014, Proceedings, Part I. Lecture Notes in Computer Science, vol. 8325, New York, NY, pp. 133-145, International Conference on Multimedia Modeling, Dublin, Ireland, 6/01/14. https://doi.org/10.1007/978-3-319-04114-8_12

Collections for automatic image annotation and photo tag recommendation. / Mcparlane, Philip J.; Moshfeghi, Yashar; Jose, Joemon M.

MultiMedia Modeling: 20th Anniversary International Conference, MMM 2014, Dublin, Ireland, January 6-10, 2014, Proceedings, Part I. New York, NY, 2013. p. 133-145 (Lecture Notes in Computer Science; Vol. 8325).

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

TY - GEN

T1 - Collections for automatic image annotation and photo tag recommendation

AU - Mcparlane, Philip J.

AU - Moshfeghi, Yashar

AU - Jose, Joemon M.

PY - 2013/12/20

Y1 - 2013/12/20

N2 - This paper highlights a number of problems which exist in the evaluation of existing image annotation and tag recommendation methods. Crucially, the collections used by these state-of-the-art methods contain a number of biases which may be exploited or detrimental to their evaluation, resulting in misleading results. In total we highlight seven issues for three popular annotation evaluation collections, i.e. Corel5k, ESP Game and IAPR, as well as three issues with collections used in two state-of-the-art photo tag recommendation methods. The result of this paper is two freely available Flickr image collections designed for the fair evaluation of image annotation and tag recommendation methods called Flickr-AIA and Flickr-PTR respectively. We show through experimentation and demonstration that these collection are ultimately fairer benchmarks than existing collections.

AB - This paper highlights a number of problems which exist in the evaluation of existing image annotation and tag recommendation methods. Crucially, the collections used by these state-of-the-art methods contain a number of biases which may be exploited or detrimental to their evaluation, resulting in misleading results. In total we highlight seven issues for three popular annotation evaluation collections, i.e. Corel5k, ESP Game and IAPR, as well as three issues with collections used in two state-of-the-art photo tag recommendation methods. The result of this paper is two freely available Flickr image collections designed for the fair evaluation of image annotation and tag recommendation methods called Flickr-AIA and Flickr-PTR respectively. We show through experimentation and demonstration that these collection are ultimately fairer benchmarks than existing collections.

KW - image annotation

KW - tag recommendation methods

KW - evaluation

KW - social media

UR - https://link.springer.com/

U2 - 10.1007/978-3-319-04114-8_12

DO - 10.1007/978-3-319-04114-8_12

M3 - Conference contribution book

SN - 9783319041131

SN - 9783319041148

T3 - Lecture Notes in Computer Science

SP - 133

EP - 145

BT - MultiMedia Modeling

CY - New York, NY

ER -

Mcparlane PJ, Moshfeghi Y, Jose JM. Collections for automatic image annotation and photo tag recommendation. In MultiMedia Modeling: 20th Anniversary International Conference, MMM 2014, Dublin, Ireland, January 6-10, 2014, Proceedings, Part I. New York, NY. 2013. p. 133-145. (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-319-04114-8_12