Case study: Mapping potential informal settlements areas in Tegucigalpa with machine learning to plan ground survey

TL;DR

This paper demonstrates how machine learning applied to satellite imagery can efficiently identify potential informal settlement areas, providing a scalable alternative to traditional census methods for urban planning in Tegucigalpa.

Contribution

It introduces a novel approach combining open data and machine learning to map informal settlements, reducing reliance on costly field surveys.

Findings

Successful identification of informal settlement areas in Tegucigalpa

Demonstrated scalability of satellite image analysis for urban monitoring

Provided a replicable methodology for similar urban contexts

Abstract

Data collection through censuses is conducted every 10 years on average in Latin America, making it difficult to monitor the growth and support needed by communities living in these settlements. Conducting a field survey requires logistical resources to be able to do it exhaustively. The increasing availability of open data, high-resolution satellite images, and free software to process them allow us to be able to do so in a scalable way based on the analysis of these sources of information. This case study shows the collaboration between Dymaxion Labs and the NGO Techo to employ machine learning techniques to create the first informal settlements census of Tegucigalpa, Honduras.