Automatic land cover classification of geotagged images using ID3, Naïve Bayes and Random Forest

M. Octaviano Pratama, Aniati Murni Arymurthy

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

Abstract

Land cover represents characteristics of earth surface. By utilizing the abundance of geotagged images from online crowdsource images like Geotagged photo library (http://eomf.ou.edu/nhotosi from the University of Oklahoma, prediction of land cover types will be established by using machine learning techniques. RGB Histogram, Edge Orientation and Vegetation Indices were used to obtain 8 features that representing images, therefore several classifiers were performed to observe which of classifiers produce best accuracy. Best classifier then used to predict unclassified images. The result, Random Forest classifier produces 82% in overall validation accuracy and 89% of 74 unclassified images was successfully predicted comparing with expert prediction result The last, 74 of successful predicted images were mapped into Geographic Information System (GIS) to show land cover in GIS. This model was measured by using precision, recall, F-Test and Kappa Coefficient The performance of each measurement reaches 89.8%, 88.1%, 88.6%, 85.6% respectively.

Original languageEnglish
Title of host publication2017 International Conference on Advanced Computer Science and Information Systems, ICACSIS 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages245-249
Number of pages5
ISBN (Electronic)9781538631720
DOIs
Publication statusPublished - 4 May 2018
Event9th International Conference on Advanced Computer Science and Information Systems, ICACSIS 2017 - Jakarta, Indonesia
Duration: 28 Oct 201729 Oct 2017

Publication series

Name2017 International Conference on Advanced Computer Science and Information Systems, ICACSIS 2017
Volume2018-January

Conference

Conference9th International Conference on Advanced Computer Science and Information Systems, ICACSIS 2017
CountryIndonesia
CityJakarta
Period28/10/1729/10/17

Keywords

  • Geotagged Image
  • ID3
  • Land Cover
  • Naïve Bayes
  • Random Forest

Fingerprint Dive into the research topics of 'Automatic land cover classification of geotagged images using ID3, Naïve Bayes and Random Forest'. Together they form a unique fingerprint.

  • Cite this

    Pratama, M. O., & Arymurthy, A. M. (2018). Automatic land cover classification of geotagged images using ID3, Naïve Bayes and Random Forest. In 2017 International Conference on Advanced Computer Science and Information Systems, ICACSIS 2017 (pp. 245-249). (2017 International Conference on Advanced Computer Science and Information Systems, ICACSIS 2017; Vol. 2018-January). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICACSIS.2017.8355041