Geostatistics for context-aware image classification

Felipe Codevilla, Silvia S.C. Botelho, Nelson Duarte, Sam Purkis, A. S.M. Shihavuddin, Rafael Garcia, Nuno Gracias

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

4 Citations (Scopus)

Abstract

Context information is fundamental for image understanding. Many algorithms add context information by including semantic relations among objects such as neighboring tendencies, relative sizes and positions. To achieve context inclusion, popular context-aware classification methods rely on probabilistic graphical models such as Markov Random Fields (MRF) or Conditional Random Fields (CRF). However, recent studies showed that MRF/CRF approaches do not perform better than a simple smoothing on the labeling results. The need for more context awareness has motivated the use of different methods where the semantic relations between objects are further enforced. With this, we found that on particular application scenarios where some specific assumptions can be made, the use of context relationships is greatly more effective. We propose a new method, called GeoSim, to compute the labels of mosaic images with context label agreement. Our method trains a transition probability model to enforce properties such as class size and proportions. The method draws inspiration from Geostatistics, usually used to model spatial uncertainties. We tested the proposed method in two different ocean seabed classification context, obtaining state-of-art results.

Original languageEnglish (US)
Title of host publicationComputer Vision Systems - 10th International Conference, ICVS 2015, Proceedings
EditorsAntonios Gasteratos, Lazaros Nalpantidis, Volker Kruger, Jan-Olof Eklundh
PublisherSpringer Verlag
Pages228-239
Number of pages12
ISBN (Print)9783319209036
DOIs
StatePublished - Jan 1 2015
Externally publishedYes
Event10th International Conference on Computer Vision Systems, ICVS 2015 - Copenhagen, Denmark
Duration: Jul 6 2015Jul 9 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9163
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other10th International Conference on Computer Vision Systems, ICVS 2015
CountryDenmark
CityCopenhagen
Period7/6/157/9/15

Fingerprint

Geostatistics
Image classification
Image Classification
Context-aware
Labels
Semantics
Image understanding
Conditional Random Fields
Labeling
Random Field
Image Mosaic
Image Understanding
Transition Model
Context-awareness
Spatial Model
Probability Model
Graphical Models
Transition Probability
Probabilistic Model
Ocean

Keywords

  • Conditional random fields
  • Context adding
  • Geostatistics
  • Underwater vision

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Codevilla, F., Botelho, S. S. C., Duarte, N., Purkis, S., Shihavuddin, A. S. M., Garcia, R., & Gracias, N. (2015). Geostatistics for context-aware image classification. In A. Gasteratos, L. Nalpantidis, V. Kruger, & J-O. Eklundh (Eds.), Computer Vision Systems - 10th International Conference, ICVS 2015, Proceedings (pp. 228-239). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9163). Springer Verlag. https://doi.org/10.1007/978-3-319-20904-3_22

Geostatistics for context-aware image classification. / Codevilla, Felipe; Botelho, Silvia S.C.; Duarte, Nelson; Purkis, Sam; Shihavuddin, A. S.M.; Garcia, Rafael; Gracias, Nuno.

Computer Vision Systems - 10th International Conference, ICVS 2015, Proceedings. ed. / Antonios Gasteratos; Lazaros Nalpantidis; Volker Kruger; Jan-Olof Eklundh. Springer Verlag, 2015. p. 228-239 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9163).

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

Codevilla, F, Botelho, SSC, Duarte, N, Purkis, S, Shihavuddin, ASM, Garcia, R & Gracias, N 2015, Geostatistics for context-aware image classification. in A Gasteratos, L Nalpantidis, V Kruger & J-O Eklundh (eds), Computer Vision Systems - 10th International Conference, ICVS 2015, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9163, Springer Verlag, pp. 228-239, 10th International Conference on Computer Vision Systems, ICVS 2015, Copenhagen, Denmark, 7/6/15. https://doi.org/10.1007/978-3-319-20904-3_22
Codevilla F, Botelho SSC, Duarte N, Purkis S, Shihavuddin ASM, Garcia R et al. Geostatistics for context-aware image classification. In Gasteratos A, Nalpantidis L, Kruger V, Eklundh J-O, editors, Computer Vision Systems - 10th International Conference, ICVS 2015, Proceedings. Springer Verlag. 2015. p. 228-239. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-20904-3_22
Codevilla, Felipe ; Botelho, Silvia S.C. ; Duarte, Nelson ; Purkis, Sam ; Shihavuddin, A. S.M. ; Garcia, Rafael ; Gracias, Nuno. / Geostatistics for context-aware image classification. Computer Vision Systems - 10th International Conference, ICVS 2015, Proceedings. editor / Antonios Gasteratos ; Lazaros Nalpantidis ; Volker Kruger ; Jan-Olof Eklundh. Springer Verlag, 2015. pp. 228-239 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{8c5b2b3e91fe4d41bfbad2d16891dfa6,
title = "Geostatistics for context-aware image classification",
abstract = "Context information is fundamental for image understanding. Many algorithms add context information by including semantic relations among objects such as neighboring tendencies, relative sizes and positions. To achieve context inclusion, popular context-aware classification methods rely on probabilistic graphical models such as Markov Random Fields (MRF) or Conditional Random Fields (CRF). However, recent studies showed that MRF/CRF approaches do not perform better than a simple smoothing on the labeling results. The need for more context awareness has motivated the use of different methods where the semantic relations between objects are further enforced. With this, we found that on particular application scenarios where some specific assumptions can be made, the use of context relationships is greatly more effective. We propose a new method, called GeoSim, to compute the labels of mosaic images with context label agreement. Our method trains a transition probability model to enforce properties such as class size and proportions. The method draws inspiration from Geostatistics, usually used to model spatial uncertainties. We tested the proposed method in two different ocean seabed classification context, obtaining state-of-art results.",
keywords = "Conditional random fields, Context adding, Geostatistics, Underwater vision",
author = "Felipe Codevilla and Botelho, {Silvia S.C.} and Nelson Duarte and Sam Purkis and Shihavuddin, {A. S.M.} and Rafael Garcia and Nuno Gracias",
year = "2015",
month = "1",
day = "1",
doi = "10.1007/978-3-319-20904-3_22",
language = "English (US)",
isbn = "9783319209036",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "228--239",
editor = "Antonios Gasteratos and Lazaros Nalpantidis and Volker Kruger and Jan-Olof Eklundh",
booktitle = "Computer Vision Systems - 10th International Conference, ICVS 2015, Proceedings",
address = "Germany",

}

TY - GEN

T1 - Geostatistics for context-aware image classification

AU - Codevilla, Felipe

AU - Botelho, Silvia S.C.

AU - Duarte, Nelson

AU - Purkis, Sam

AU - Shihavuddin, A. S.M.

AU - Garcia, Rafael

AU - Gracias, Nuno

PY - 2015/1/1

Y1 - 2015/1/1

N2 - Context information is fundamental for image understanding. Many algorithms add context information by including semantic relations among objects such as neighboring tendencies, relative sizes and positions. To achieve context inclusion, popular context-aware classification methods rely on probabilistic graphical models such as Markov Random Fields (MRF) or Conditional Random Fields (CRF). However, recent studies showed that MRF/CRF approaches do not perform better than a simple smoothing on the labeling results. The need for more context awareness has motivated the use of different methods where the semantic relations between objects are further enforced. With this, we found that on particular application scenarios where some specific assumptions can be made, the use of context relationships is greatly more effective. We propose a new method, called GeoSim, to compute the labels of mosaic images with context label agreement. Our method trains a transition probability model to enforce properties such as class size and proportions. The method draws inspiration from Geostatistics, usually used to model spatial uncertainties. We tested the proposed method in two different ocean seabed classification context, obtaining state-of-art results.

AB - Context information is fundamental for image understanding. Many algorithms add context information by including semantic relations among objects such as neighboring tendencies, relative sizes and positions. To achieve context inclusion, popular context-aware classification methods rely on probabilistic graphical models such as Markov Random Fields (MRF) or Conditional Random Fields (CRF). However, recent studies showed that MRF/CRF approaches do not perform better than a simple smoothing on the labeling results. The need for more context awareness has motivated the use of different methods where the semantic relations between objects are further enforced. With this, we found that on particular application scenarios where some specific assumptions can be made, the use of context relationships is greatly more effective. We propose a new method, called GeoSim, to compute the labels of mosaic images with context label agreement. Our method trains a transition probability model to enforce properties such as class size and proportions. The method draws inspiration from Geostatistics, usually used to model spatial uncertainties. We tested the proposed method in two different ocean seabed classification context, obtaining state-of-art results.

KW - Conditional random fields

KW - Context adding

KW - Geostatistics

KW - Underwater vision

UR - http://www.scopus.com/inward/record.url?scp=84948956581&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84948956581&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-20904-3_22

DO - 10.1007/978-3-319-20904-3_22

M3 - Conference contribution

SN - 9783319209036

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 228

EP - 239

BT - Computer Vision Systems - 10th International Conference, ICVS 2015, Proceedings

A2 - Gasteratos, Antonios

A2 - Nalpantidis, Lazaros

A2 - Kruger, Volker

A2 - Eklundh, Jan-Olof

PB - Springer Verlag

ER -