Exploring the potential of machine learning for automatic slum identification from VHR imagery

J.C. Duque; J.E. Patino; A. Betancourt

doi:10.3390/rs9090895

Exploring the potential of machine learning for automatic slum identification from VHR imagery

J.C. Duque
, J.E. Patino
, A. Betancourt

Research output: Contribution to journal › Article › Academic › peer-review

100 Citations (Scopus)

385 Downloads (Pure)

Abstract

Slum identification in urban settlements is a crucial step in the process of formulation of pro-poor policies. However, the use of conventional methods for slum detection such as field surveys can be time-consuming and costly. This paper explores the possibility of implementing a low-cost standardized method for slum detection. We use spectral, texture and structural features extracted from very high spatial resolution imagery as input data and evaluate the capability of three machine learning algorithms (Logistic Regression, Support Vector Machine and Random Forest) to classify urban areas as slum or no-slum. Using data from Buenos Aires (Argentina), Medellin (Colombia) and Recife (Brazil), we found that Support Vector Machine with radial basis kernel delivers the best performance (with F2-scores over 0.81). We also found that singularities within cities preclude the use of a unified classification model.

Original language	English
Article number	895
Number of pages	13
Journal	Remote Sensing
Volume	9
Issue number	9
DOIs	https://doi.org/10.3390/rs9090895
Publication status	Published - 1 Sept 2017

Funding

Acknowledgments: We gratefully acknowledge the financial support from CAF—Development Bank of Latin America. We also thank to the Google Company for providing freely images for research purpose. Finally, we thank the anonymous Remote Sensing reviewers for their insightful and helpful comments during the review process. The usual disclaimer applies.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 11 Sustainable Cities and Communities

Keywords

Machine learning
Remote sensing
Slum detection

Access to Document

10.3390/rs9090895

Publishers versionFinal published version, 7.55 MB

Cite this

@article{a012305f35ff4876a5daffb98ac799c3,

title = "Exploring the potential of machine learning for automatic slum identification from VHR imagery",

abstract = "Slum identification in urban settlements is a crucial step in the process of formulation of pro-poor policies. However, the use of conventional methods for slum detection such as field surveys can be time-consuming and costly. This paper explores the possibility of implementing a low-cost standardized method for slum detection. We use spectral, texture and structural features extracted from very high spatial resolution imagery as input data and evaluate the capability of three machine learning algorithms (Logistic Regression, Support Vector Machine and Random Forest) to classify urban areas as slum or no-slum. Using data from Buenos Aires (Argentina), Medellin (Colombia) and Recife (Brazil), we found that Support Vector Machine with radial basis kernel delivers the best performance (with F2-scores over 0.81). We also found that singularities within cities preclude the use of a unified classification model.",

keywords = "Machine learning, Remote sensing, Slum detection",

author = "J.C. Duque and J.E. Patino and A. Betancourt",

year = "2017",

month = sep,

day = "1",

doi = "10.3390/rs9090895",

language = "English",

volume = "9",

journal = "Remote Sensing",

publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",

number = "9",

}

TY - JOUR

T1 - Exploring the potential of machine learning for automatic slum identification from VHR imagery

AU - Duque, J.C.

AU - Patino, J.E.

AU - Betancourt, A.

PY - 2017/9/1

Y1 - 2017/9/1

N2 - Slum identification in urban settlements is a crucial step in the process of formulation of pro-poor policies. However, the use of conventional methods for slum detection such as field surveys can be time-consuming and costly. This paper explores the possibility of implementing a low-cost standardized method for slum detection. We use spectral, texture and structural features extracted from very high spatial resolution imagery as input data and evaluate the capability of three machine learning algorithms (Logistic Regression, Support Vector Machine and Random Forest) to classify urban areas as slum or no-slum. Using data from Buenos Aires (Argentina), Medellin (Colombia) and Recife (Brazil), we found that Support Vector Machine with radial basis kernel delivers the best performance (with F2-scores over 0.81). We also found that singularities within cities preclude the use of a unified classification model.

AB - Slum identification in urban settlements is a crucial step in the process of formulation of pro-poor policies. However, the use of conventional methods for slum detection such as field surveys can be time-consuming and costly. This paper explores the possibility of implementing a low-cost standardized method for slum detection. We use spectral, texture and structural features extracted from very high spatial resolution imagery as input data and evaluate the capability of three machine learning algorithms (Logistic Regression, Support Vector Machine and Random Forest) to classify urban areas as slum or no-slum. Using data from Buenos Aires (Argentina), Medellin (Colombia) and Recife (Brazil), we found that Support Vector Machine with radial basis kernel delivers the best performance (with F2-scores over 0.81). We also found that singularities within cities preclude the use of a unified classification model.

KW - Machine learning

KW - Remote sensing

KW - Slum detection

UR - https://www.scopus.com/pages/publications/85029393426

U2 - 10.3390/rs9090895

DO - 10.3390/rs9090895

M3 - Article

AN - SCOPUS:85029393426

VL - 9

JO - Remote Sensing

JF - Remote Sensing

IS - 9

M1 - 895

ER -

Exploring the potential of machine learning for automatic slum identification from VHR imagery

Abstract

Funding

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this