Aggregated deep local features for remote sensing image retrieval

Research output: Contribution to journalSpecial issueAcademicpeer-review

1 Citation (Scopus)

Abstract

Remote Sensing Image Retrieval remains a challenging topic due to the special nature of Remote Sensing imagery. Such images contain various different semantic objects, which clearly complicates the retrieval task. In this paper, we present an image retrieval pipeline that uses attentive, local convolutional features and aggregates them using the Vector of Locally Aggregated Descriptors (VLAD) to produce a global descriptor. We study various system parameters such as the multiplicative and additive attention mechanisms and descriptor dimensionality. We propose a query expansion method that requires no external inputs. Experiments demonstrate that even without training, the local convolutional features and global representation outperform other systems. After system tuning, we can achieve state-of-the-art or competitive results. Furthermore, we observe that our query expansion method increases overall system performance by about 3%, using only the top-three retrieved images. Finally, we show how dimensionality reduction produces compact descriptors with increased retrieval performance and fast retrieval computation times, e.g., 50% faster than the current systems.
LanguageEnglish
Article number493
Number of pages23
JournalRemote Sensing
Volume11
Issue number5
DOIs
StatePublished - 28 Feb 2019

Fingerprint

remote sensing
imagery
experiment
method
state of the art
parameter

Keywords

  • image retrieval
  • convolutional descriptor
  • query expansion
  • descriptor aggregation
  • Convolutional Neural Networks

Cite this

@article{473949e7fe6942e3b1cfdd2d341f19c4,
title = "Aggregated deep local features for remote sensing image retrieval",
abstract = "Remote Sensing Image Retrieval remains a challenging topic due to the special nature of Remote Sensing imagery. Such images contain various different semantic objects, which clearly complicates the retrieval task. In this paper, we present an image retrieval pipeline that uses attentive, local convolutional features and aggregates them using the Vector of Locally Aggregated Descriptors (VLAD) to produce a global descriptor. We study various system parameters such as the multiplicative and additive attention mechanisms and descriptor dimensionality. We propose a query expansion method that requires no external inputs. Experiments demonstrate that even without training, the local convolutional features and global representation outperform other systems. After system tuning, we can achieve state-of-the-art or competitive results. Furthermore, we observe that our query expansion method increases overall system performance by about 3{\%}, using only the top-three retrieved images. Finally, we show how dimensionality reduction produces compact descriptors with increased retrieval performance and fast retrieval computation times, e.g., 50{\%} faster than the current systems.",
keywords = "image retrieval, convolutional descriptor, query expansion, descriptor aggregation, Convolutional Neural Networks",
author = "Raffaele Imbriaco and Clint Sebastian and Egor Bondarau and {de With}, Peter",
year = "2019",
month = "2",
day = "28",
doi = "10.3390/rs11050493",
language = "English",
volume = "11",
journal = "Remote Sensing",
issn = "2072-4292",
publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",
number = "5",

}

Aggregated deep local features for remote sensing image retrieval. / Imbriaco, Raffaele (Corresponding author); Sebastian, Clint; Bondarau, Egor; de With, Peter.

In: Remote Sensing, Vol. 11, No. 5, 493, 28.02.2019.

Research output: Contribution to journalSpecial issueAcademicpeer-review

TY - JOUR

T1 - Aggregated deep local features for remote sensing image retrieval

AU - Imbriaco,Raffaele

AU - Sebastian,Clint

AU - Bondarau,Egor

AU - de With,Peter

PY - 2019/2/28

Y1 - 2019/2/28

N2 - Remote Sensing Image Retrieval remains a challenging topic due to the special nature of Remote Sensing imagery. Such images contain various different semantic objects, which clearly complicates the retrieval task. In this paper, we present an image retrieval pipeline that uses attentive, local convolutional features and aggregates them using the Vector of Locally Aggregated Descriptors (VLAD) to produce a global descriptor. We study various system parameters such as the multiplicative and additive attention mechanisms and descriptor dimensionality. We propose a query expansion method that requires no external inputs. Experiments demonstrate that even without training, the local convolutional features and global representation outperform other systems. After system tuning, we can achieve state-of-the-art or competitive results. Furthermore, we observe that our query expansion method increases overall system performance by about 3%, using only the top-three retrieved images. Finally, we show how dimensionality reduction produces compact descriptors with increased retrieval performance and fast retrieval computation times, e.g., 50% faster than the current systems.

AB - Remote Sensing Image Retrieval remains a challenging topic due to the special nature of Remote Sensing imagery. Such images contain various different semantic objects, which clearly complicates the retrieval task. In this paper, we present an image retrieval pipeline that uses attentive, local convolutional features and aggregates them using the Vector of Locally Aggregated Descriptors (VLAD) to produce a global descriptor. We study various system parameters such as the multiplicative and additive attention mechanisms and descriptor dimensionality. We propose a query expansion method that requires no external inputs. Experiments demonstrate that even without training, the local convolutional features and global representation outperform other systems. After system tuning, we can achieve state-of-the-art or competitive results. Furthermore, we observe that our query expansion method increases overall system performance by about 3%, using only the top-three retrieved images. Finally, we show how dimensionality reduction produces compact descriptors with increased retrieval performance and fast retrieval computation times, e.g., 50% faster than the current systems.

KW - image retrieval

KW - convolutional descriptor

KW - query expansion

KW - descriptor aggregation

KW - Convolutional Neural Networks

U2 - 10.3390/rs11050493

DO - 10.3390/rs11050493

M3 - Special issue

VL - 11

JO - Remote Sensing

T2 - Remote Sensing

JF - Remote Sensing

SN - 2072-4292

IS - 5

M1 - 493

ER -