Across Africa, an estimated 62 per cent of urban dwellings are informal, and by 2050, its urban population is projected to double from 700 million to 1.4 billion, meaning two-thirds of Africans will live in cities. While the continent continues to experience the fastest urban growth worldwide, the gap between housing demand and infrastructure is deepening. As a result, many communities lack adequate access to water, sanitation, electricity, waste management systems and reliable public transport.
In South Africa, municipalities receive funding by the national government for the provision of free basic services based on the number of households within each municipal area. Reliable household counts are therefore essential for designing policy interventions, planning infrastructure, and upgrading underserved neighbourhoods. For municipalities such as eThekwini, ensuring the accuracy of these figures is critical for directing resources to disadvantaged communities.
UNITAC Hamburg first partnered with eThekwini Municipality in 2021 and 2022, through an Open Call, as part of a project funded by the German Federal Foreign Office. The aim was to explore how innovative technologies could support equitable service delivery. The initiative aimed to automate and accelerate land-mapping processes, improve data accessibility and strengthen the city’s capacity to prioritize upgrading interventions in underserved communities. It also sought to provide updated records and insights on informal settlement dynamics. Before commencement, the partners outlined a set of principles to guide the project, ensuring responsible data use, the protection of human rights and safeguarding vulnerable communities.
A key output of the collaboration was the Building & Establishment Automated Mapper (BEAM), an innovative tool that uses machine learning to identify and map building footprints from high-resolution aerial and satellite imagery. Prior to its introduction, manually mapping informal settlements across the municipal area could take several months. With BEAM, this process was reduced to approximately 72 hours, significantly improving the timeliness of spatial data available to the municipality.
The initial version of BEAM focused primarily on dense informal settlements. However, it had limitations in detecting “backyard structures” within formal neighbourhoods. These small rental units—often shacks, wendy houses or basic brick rooms—are built on formal residential plots and typically do not have adequate access to basic services. As “backyarding” becomes an increasingly common form of accommodation in South Africa, these structures are assumed to represent a substantial share of informal housing. Yet, because they are dispersed across the city and often resemble formal buildings from above, they can be difficult to identify through satellite imagery.
While BEAM significantly improved the monitoring of informal settlement expansion, complex urban environments where formal and informal structures coexist closely remained challenging. Small or partially obscured buildings, or those with visual similarities to formal housing, were sometimes misclassified or overlooked. Such gaps can result in the under-representation of specific vulnerable groups in municipal records.
To address this challenge, the Open Call project focused on expanding BEAM’s functionality to better distinguish formal structures from informal ones. The upgrade placed particular emphasis on detecting backyard structures in planned neighbourhoods, thereby contributing to a more accurate assessment of indigent households within eThekwini’s boundaries.
To train the enhanced model, additional information on informal backyard units and rural building clusters was collected from aerial photographs and municipal GIS data. These materials were meticulously reviewed and labelled to create a high-quality dataset that taught the system to recognise different building types. This training set was validated by municipal officials to ensure that it reflected real conditions on the ground. The machine learning model was then trained using this dataset, enabling BEAM to better differentiate between formal and informal dwellings.
As the project concluded in December this year, UNITAC Hamburg handed over the upgraded version of BEAM, along with user manuals and technical documentation, ensuring that the municipality can operate, test, verify and maintain the system independently. Training sessions will further build local capacity to use BEAM effectively within eThekwini’s planning processes.
More accurate and comprehensive data on informal settlements and backyard housing will support the municipality in updating its indigent household count records, ensuring that funding for basic services can be received and allocated to previously uncounted communities. Ultimately, the updated records will help inform better planning and upgrading interventions, enabling residents across the city to benefit from enhanced infrastructure and improved access to essential services and advance inclusive, resilient and sustainable urban development in eThekwini.
Publishing date: 20 December 2025
Sources:
Department of Statistics South Africa