SciELO - Scientific Electronic Library Online

vol.113 issue4 author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand



Related links

  • On index processCited by Google
  • On index processSimilars in Google


SAIEE Africa Research Journal

On-line version ISSN 1991-1696
Print version ISSN 0038-2221


UWIZERA, Davy K.; RURANGA, Charles  and  MCSHARRY, Patrick. Classifying Economic Areas for Urban Planning using Deep Learning and Satellite Imagery in East Africa. SAIEE ARJ [online]. 2022, vol.113, n.4, pp.138-150. ISSN 1991-1696.

Monitoring and assessing the distribution of economic areas in East Africa such as low and high income neighborhoods, has typically relied on the use of structured data and traditional survey approaches for collecting information such as questionnaires, interviews and field visits. These types of surveys are slow, costly and prone to human error. With the digital revolution, a lot of unstructured data is generated daily that is likely to contain useful proxy data for many economic variables. In this research we focus on satellite imagery data with applications in East Africa. Recently East African cities have been developing at a fast pace by building new infrastructure and constructing innovative economic zones. Moreover with increased urban population, cities have been expanding in multiple directions affecting the overall distribution of areas with economic activity. Automatic detection and classification of these areas could be used to inform a number of policies such as land usage and could also assist with policy enforcement monitoring. On the other hand, the distribution of different economic areas in a specific city could provide proxies for various economic development variables such as income distribution and poverty metrics. In this research, we apply deep learning techniques to satellite imagery to classify and assess the distribution of various economic areas of a specific region for urban planning. By benchmarking performance against various state-of-art models, results show that the proposed deep learning techniques yielded superior performance with an f1-score of 99%.

Keywords : classification; deep learning; satellite imagery; transfer learning; urban planning; remote sensing; monitoring.

        · text in English     · English ( pdf )


Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License