CoPhIR Image Collection under the Microscope

D 2009

CoPhIR Image Collection under the Microscope

BATKO, Michal, Petra BUDÍKOVÁ and David NOVÁK

Basic information

Original name

CoPhIR Image Collection under the Microscope

Name in Czech

Kolekce obrázků CoPhIR pod drobnohledem

Authors

BATKO, Michal (203 Czech Republic, belonging to the institution), Petra BUDÍKOVÁ (203 Czech Republic, belonging to the institution) and David NOVÁK (203 Czech Republic, guarantor, belonging to the institution)

Edition

Washington, DC, USA, Proceedings of the 2009 Second International Workshop on Similarity Search and Applications, p. 47-54, 8 pp. 2009

Publisher

IEEE Computer Society

Other information

Language

English

Type of outcome

Stať ve sborníku

Field of Study

10201 Computer sciences, information science, bioinformatics

Country of publisher

United States of America

Confidentiality degree

není předmětem státního či obchodního tajemství

Publication form

printed version "print"

RIV identification code

RIV/00216224:14330/09:00029662

Organization unit

Faculty of Informatics

ISBN

978-0-7695-3765-8

UT WoS

000282087600006

Keywords in English

metric space; MPEG-7; visual descriptors; CoPhIR dataset; dataset analysis

Abstract

ORIG CZ

V originále

The Content-based Photo Image Retrieval (CoPhIR) dataset is the largest available database of digital images with corresponding visual descriptors. It contains five MPEG-7 global descriptors extracted from more than 106 million images from Flickr photo-sharing system. In this paper, we analyze this dataset focusing on 1) efficiency of similarity-based indexing and searching and on 2) expressiveness of combination of the descriptors with respect to subjective perception of visual similarity. We treat the descriptors as metric spaces and then combine them into a multi-metric space. We analyze distance distributions of individual descriptors, measure intrinsic dimensionality of these datasets and statistically evaluate correlation between these descriptors. Further, we use two methods to assess subjective accuracy and satisfaction of similarity retrieval based on a combination of descriptors that is recommended for CoPhIR, and we compare these results on databases of 10 and 100 million CoPhIR images. Finally, we suggest, explore and evaluate two approaches to improve the accuracy: 1) applying logarithms in order to weaken influence of a single descriptor contribution if it deviates from the rest, and 2) the possibility of categorization of the dataset and identifying visual characteristics important for individual categories.

In Czech

CoPhIR (Content-based Photo Image Retrieval) je největší dostupná databáze...

Links

GA201/09/0683, research and development project

Name: Vyhledávání v rozsáhlých multimediálních databázích

Investor: Czech Science Foundation, Similarity Searching in Very Large Multimedia Databases

Citovat

BATKO, Michal, Petra BUDÍKOVÁ and David NOVÁK. CoPhIR Image Collection under the Microscope. In Proceedings of the 2009 Second International Workshop on Similarity Search and Applications. Washington, DC, USA: IEEE Computer Society, 2009, p. 47-54. ISBN 978-0-7695-3765-8.

@inproceedings{858707,
   author = {Batko, Michal and Budíková, Petra and Novák, David},
   address = {Washington, DC, USA},
   booktitle = {Proceedings of the 2009 Second International Workshop on Similarity Search and Applications},
   keywords = {metric space; MPEG-7; visual descriptors; CoPhIR dataset; dataset analysis},
   howpublished = {tištěná verze "print"},
   language = {eng},
   location = {Washington, DC, USA},
   isbn = {978-0-7695-3765-8},
   pages = {47-54},
   publisher = {IEEE Computer Society},
   title = {CoPhIR Image Collection under the Microscope},
   year = {2009}
}

TY  - JOUR
ID  - 858707
AU  - Batko, Michal - Budíková, Petra - Novák, David
PY  - 2009
TI  - CoPhIR Image Collection under the Microscope
PB  - IEEE Computer Society
CY  - Washington, DC, USA
SN  - 9780769537658
KW  - metric space
KW  - MPEG-7
KW  - visual descriptors
KW  - CoPhIR dataset
KW  - dataset analysis
N2  - The Content-based Photo Image Retrieval (CoPhIR) dataset is the largest available database of digital images with corresponding visual descriptors. It contains five MPEG-7 global descriptors extracted from more than 106 million images from Flickr photo-sharing system. In this paper, we analyze this dataset focusing on 1) efficiency of similarity-based indexing and searching and on 2) expressiveness of combination of the descriptors with respect to subjective perception of visual similarity. We treat the descriptors as metric spaces and then combine them into a multi-metric space. We analyze distance distributions of individual descriptors, measure intrinsic dimensionality of these datasets and statistically evaluate correlation between these descriptors. Further, we use two methods to assess subjective accuracy and satisfaction of similarity retrieval based on a combination of descriptors that is recommended for CoPhIR, and we compare these results on databases of 10 and 100 million CoPhIR images. Finally, we suggest, explore and evaluate two approaches to improve the accuracy: 1) applying logarithms in order to weaken influence of a single descriptor contribution if it deviates from the rest, and 2) the possibility of categorization of the dataset and identifying visual characteristics important for individual categories.
ER  -

BATKO, Michal, Petra BUDÍKOVÁ and David NOVÁK. CoPhIR Image Collection under the Microscope. In \textit{Proceedings of the 2009 Second International Workshop on Similarity Search and Applications}. Washington, DC, USA: IEEE Computer Society, 2009, p.~47-54. ISBN~978-0-7695-3765-8.

Detailed Information on Publication Record