On very large scale test collection for landmark image search benchmarking

Zhiyong Cheng, Jialie Shen

Research output: Contribution to journalArticleResearchpeer-review

Abstract

High quality test collections have been becoming more and more important for the technological advancement in geo-referenced image retrieval and analytics. In this paper, we present a large scale test collection to support robust performance evaluation of landmark image search and corresponding construction methodology. Using the approach, we develop a very large scale test collection consisting of three key components: (1) 355,141 images of 128 landmarks in five cities across three continents crawled from Flickr; (2) different kinds of textual features for each image, including surrounding text (e.g. tags), contextual data (e.g. geo-location and upload time), and metadata (e.g. uploader and EXIF); and (3) six types of low-level visual features. In order to support robust and effective performance assessment, a series of baseline experimental studies have been conducted on the search performance over both textual and visual queries. The results demonstrate importance and effectiveness of the test collection.

Original languageEnglish
Pages (from-to)13-26
Number of pages14
JournalSignal Processing
Volume124
Early online date20 Nov 2015
DOIs
Publication statusPublished - Jul 2016
Externally publishedYes

Fingerprint

Image retrieval
Benchmarking
Metadata

Cite this

Cheng, Zhiyong ; Shen, Jialie. / On very large scale test collection for landmark image search benchmarking. In: Signal Processing. 2016 ; Vol. 124. pp. 13-26.
@article{4eca50b6bb3d4500a009e1075b1af15a,
title = "On very large scale test collection for landmark image search benchmarking",
abstract = "High quality test collections have been becoming more and more important for the technological advancement in geo-referenced image retrieval and analytics. In this paper, we present a large scale test collection to support robust performance evaluation of landmark image search and corresponding construction methodology. Using the approach, we develop a very large scale test collection consisting of three key components: (1) 355,141 images of 128 landmarks in five cities across three continents crawled from Flickr; (2) different kinds of textual features for each image, including surrounding text (e.g. tags), contextual data (e.g. geo-location and upload time), and metadata (e.g. uploader and EXIF); and (3) six types of low-level visual features. In order to support robust and effective performance assessment, a series of baseline experimental studies have been conducted on the search performance over both textual and visual queries. The results demonstrate importance and effectiveness of the test collection.",
keywords = "Large scale landmark image search, Performance evaluation",
author = "Zhiyong Cheng and Jialie Shen",
year = "2016",
month = "7",
doi = "10.1016/j.sigpro.2015.10.037",
language = "English",
volume = "124",
pages = "13--26",
journal = "Signal Processing",
issn = "0165-1684",
publisher = "Elsevier",

}

On very large scale test collection for landmark image search benchmarking. / Cheng, Zhiyong; Shen, Jialie.

In: Signal Processing, Vol. 124, 07.2016, p. 13-26.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - On very large scale test collection for landmark image search benchmarking

AU - Cheng, Zhiyong

AU - Shen, Jialie

PY - 2016/7

Y1 - 2016/7

N2 - High quality test collections have been becoming more and more important for the technological advancement in geo-referenced image retrieval and analytics. In this paper, we present a large scale test collection to support robust performance evaluation of landmark image search and corresponding construction methodology. Using the approach, we develop a very large scale test collection consisting of three key components: (1) 355,141 images of 128 landmarks in five cities across three continents crawled from Flickr; (2) different kinds of textual features for each image, including surrounding text (e.g. tags), contextual data (e.g. geo-location and upload time), and metadata (e.g. uploader and EXIF); and (3) six types of low-level visual features. In order to support robust and effective performance assessment, a series of baseline experimental studies have been conducted on the search performance over both textual and visual queries. The results demonstrate importance and effectiveness of the test collection.

AB - High quality test collections have been becoming more and more important for the technological advancement in geo-referenced image retrieval and analytics. In this paper, we present a large scale test collection to support robust performance evaluation of landmark image search and corresponding construction methodology. Using the approach, we develop a very large scale test collection consisting of three key components: (1) 355,141 images of 128 landmarks in five cities across three continents crawled from Flickr; (2) different kinds of textual features for each image, including surrounding text (e.g. tags), contextual data (e.g. geo-location and upload time), and metadata (e.g. uploader and EXIF); and (3) six types of low-level visual features. In order to support robust and effective performance assessment, a series of baseline experimental studies have been conducted on the search performance over both textual and visual queries. The results demonstrate importance and effectiveness of the test collection.

KW - Large scale landmark image search

KW - Performance evaluation

UR - http://www.scopus.com/inward/record.url?scp=84960229929&partnerID=8YFLogxK

U2 - 10.1016/j.sigpro.2015.10.037

DO - 10.1016/j.sigpro.2015.10.037

M3 - Article

VL - 124

SP - 13

EP - 26

JO - Signal Processing

JF - Signal Processing

SN - 0165-1684

ER -