Cities are contradictory human constructions: although they offer opportunities, they also harbor deep miseries. These inequalities are located in specific physical zones, commonly known as slums (informal human settlements). With nearly a quarter of the world’s urban population living in them, acting to stop millions of people from living in highly vulnerable conditions is a matter of global social justice. Thus, to anticipate crises and design citizen-centered responses, it is essential to generate reliable data with global coverage.
When identifying poverty in cities, socioeconomic indicators such as income, literacy or housing conditions are often used, but the physical characteristics of the place are overlooked. The differences within the city must be taken into account because they lead to greater vulnerability and risk of exclusion, exposing the inhabitants to other facets of poverty that not only encompass the monetary dimension. For example, an area prone to floods or fires makes its inhabitants more exposed to health risks and this also results in greater economic fragility.
Currently, more than sixty percent of the African population lives in slums and it is estimated that this will triple in less than thirty years
Identifying the spatial patterns that characterize urban poverty is essential to guide the development of strategies aimed at fighting for more inclusive cities. First, it can help local policies to detect the most critical areas and guide urban regeneration programs. And second, to the global ones, to generate quality and up-to-date open data that measure, through indicators, the Sustainable Development Goals (SDG) of the United Nations. The data would help launch integrated action plans for equitable global advancement and end poverty (SDG 1 and SDG 11).
However, detailed and disaggregated data at the intra-urban scale are still lacking. This prevents measuring and characterizing the physical differences that help to understand the spatial and temporal differences in living conditions in cities. Censuses and surveys often provide urban data on households, such as housing characteristics and the socioeconomic status of the inhabitants, but their scant periodicity and gaps in coverage remain inconsistent when it comes to slums. This make them ineffective tools for tackling poverty at its roots.
Earth Observation Science provides geolocated data resources, also called geodata, such as remote sensing (for example, satellite imagery), with full coverage to characterize poverty and fill data gaps both locally and across borders. global scale. Artificial intelligence, through machine learning techniques, makes it possible to systematically analyze satellite images and create efficient and transferable processes to capture the characteristics of the physical environment.
The number of poverty studies based on remote sensing has increased in the last decade, highlighting the ability to locate poor neighborhoods with greater profitability, coverage, detail and frequency than traditional methods such as censuses or surveys. Most of the studies focus on mapping the extension and its location, drawing the limits of these and opposing them to the rest of the city. However, urban poverty is not a mere binary phenomenon, that is, slum versus not slum; There are levels of poverty among slums and other planned areas of the city, as well as within each of these. For example, there are differences in the type of construction, in the proximity to risk areas such as landfills or floodable rivers, in accessibility to urban services such as schools or hospitals.
Urban poverty is not a mere binary phenomenon; levels of poverty exist between slums and other planned areas of the city, as well as within each of these
Through high-resolution satellite images and machine learning techniques, it has become clear that there are large intra-physical differences in neighborhoods with higher poverty rates. For example, an artificial intelligence model has managed to extract from satellite images of various slums in African cities the outline of various urban elements such as buildings, trees, ground surface, rivers, rubbish bins and cars. Morphological metrics have also been applied and a great diversity in its physical constitution has been detected, characterized by the size of the buildings, the type of width of the streets, the distinctive internal irregularity of each structure and the orientation patterns that make up its whole. . If this work were done by different people, it would take hundreds of hours and lots of errors, while the algorithm does it very precisely in a matter of seconds.
In sub-Saharan Africa, this line of research that combines artificial intelligence and satellite images is going to be very promising, since the population and levels of urbanization are growing in these territories –and it is expected that they will continue to do so– at uncontrollable levels. Currently, more than 70% of the African population lives in slums and it is estimated that it will triple in less than 30 years, reaching the continent to house more than two billion citizens in conditions of vulnerability. It is time to look for creative and effective methods to reverse the situation.
You can follow PLANETA FUTURO at Twitter, Facebook and Instagramand subscribe here to our ‘newsletter’.
#Artificial #intelligence #construction #inclusive #cities