Estimating poverty using aerial images: South African application

Year 2018, Volume: 1 Issue: 1, 29 - 36, 26.12.2018

Vongani H. Maluleke Sebnem Er Quentin R. Williams

Abstract

Policy makers and the government rely heavily on survey data when making policy-related decisions. Survey data is labour intensive, costly and time consuming, hence it cannot be frequently or extensively collected. The main aim of this research is to demonstrate how deep learning in computer vision coupled with statistical regression modelling can be used to estimate poverty on aerial images supplemented with national household survey data. This is executed in two phases; aerial classification and detection phase and poverty modelling phase. The aerial classification and detection phase use convolutional neural networks (CNN) to perform settlement typology classification of the aerial images into three broad geo-type classes namely; urban, rural and farm. This is then followed by object detection to detect three broad dwelling type classes in the aerial images namely; brick house, traditional house, and informal settlement. Mask Region-based CNN (Mask R-CNN) model with a resnet101 backbone model is used to perform this task. The second phase, poverty modelling phase, involves using National Income Dynamics Survey (NIDS) data to compute the poverty measure Sen-Shorrocks-Thon index (SST). This is followed by using ridge regression to model the poverty measure using aggregated results from the aerial classification and detection phase. The study area for this research is eThekwini district in Kwa-Zulu Natal, South Africa. However, this approach can be extended to other districts in South Africa.

Keywords

Poverty, Computer vision, statistical regression modelling, Aerial Image(s).

References

• [1] D. Narayan, R. Patel, K. Schafft, A. Rademacher ve S. Koch-Schulte, Voices of the poor: can anyone hear us?. New York: Oxford University Press, 2000. [Online]. Available: http://documents.worldbank.org/curated/en/131441468779067441/Voices-of-the-poor-can-anyone-hear-us.
• [2] National Planning Commission, Our future-make it works, 2011.
• [3] H. Bhorat, M. Leibbrandt, M. Maziya, S. van der Berg ve I. Woolard, Fighting Poverty-Labour Markets and Inequality in South Africa. UCT Press, 2001.
• [4] M. Xie, N. Jean, M. Burke, D. Lobell ve S. Ermon, “Transfer Learning from Deep Features for Remote Sensing and Poverty Mapping,” 2015. CoRR abs/1510.00098. http://arxiv.org/abs/1510.00098.
• [5] R. Engstrom, S. H. Jonathan ve D. L. Newhouse, “Poverty from Space: Using High Resolution Satellite Imagery for Estimating Economic Well-being”. Washington, D.C: World Bank Group, 2017. [Online]. Available: http://documents.worldbank.org/curated/en/610771513691888412/Poverty-from-space-using-high-resolution-satellite-imagery-for-estimating-economic-well-being.
• [6] Statistics South Africa, National Poverty Lines. Technical report, Pretoria: Statistics South Africa, 2018.
• [7] J. Haughton ve R. Khandker, Handbook on Poverty and Inequality. Washington, DC: The World Bank, 2009.
• [8] P. Govender, N. Kambaran, N. Patchett, A. Ruddle, G. Torr ve N. Zyl, “Poverty and Inequality in South Africa and the world,”. doi:10.4314/saaj.v7i1.24511, 2007.
• [9] A. Sen, “Poverty: An Ordinal Aproach to Measurement,” Econometrica, ss. 219-231, 1976.
• [10] I. Goodfellow, Y. Bengio ve A. Courville, Deep Learning. MIT Press, 2016. [Online]. Available: http://www.deeplearningbook.org.
• [11] M. D. Zeiler ve R. Fergus, “Visualizing and Understanding Convolutional Networks.” CoRR abs/1311.2901, 2013. [Online]. Available: http://arxiv.org/abs/1311.2901.
• [12] K. He, G. Georgia, P. Dollar ve R. Girshick, “Mask R-CNN.” CoRR abs/1703.06870, 2017. [Online]. Available: http://arxiv.org/abs/1703.06870.
• [13] G. James, D. Witten, T. Hastie ve R. Tibshirani, An Introduction to Statistical Learning with Applications in R. Springer New York Heidelberg Dordrecht, 2013.
• [14] A. E. Hoerl ve R. W. Kennard, “Ridge Regression: Biased Estimation for Problems Nonorthogonal.” Technometrics, ss. 55-67.
• [15] Southern Africa Labour and Development Research Unit, “National Income Dynamics Study 2014 -2015, Wave 4 [dataset]. Version 1.1.” Cape Town: Southern Africa Labour and Development Research Unit [producer]. (Cape Town: DataFirst [distributor], 2016. Pretoria: Department of Planning Monitoring and Evaluation [commissioner], 2014), 1970.
• [16] University Michigan State, n.d. South Africa 1:50,000. [Online]. Available: https://lib.msu.edu/branches/map/findingaids/SouthAfrica50k_KZN/.
• [17] Tzutalin, labelimg, 2015.
• [18] V. Kshirsagar, J. Wieczorek, S. Ramanathan ve R. Wells, “Household poverty classification in data-scarce environments: a machine learning approach,” 2017. arXiv preprint arXiv:1711.06813.