Access to clean water and sanitation is essential for health and sustainable development, but many African countries lack the resources and data needed to track where infrastructure is available and where communities are underserved. Traditional monitoring relies on field surveys and government reporting, which can be expensive, infrequent, and inconsistent across regions.

In this study, we developed an AI‑enabled framework that combines satellite imagery, household survey data, and population maps to provide continent‑wide, high‑resolution estimates of access to piped water and sanitation. Using Vision Transformers, a type of computer vision model, our approach can detect patterns in the environment and built infrastructure that are associated with water and sanitation services.

The framework produces national and subnational maps of infrastructure access, highlights underserved areas, and aligns with Sustainable Development Goal 6. This method provides a scalable, cost‑effective way to support water resource planning in data‑limited regions.