UNHCR Uses AI and Drones to Map Refugee Settlements in Kenya

The UN Refugee Agency (UNHCR) has successfully deployed artificial intelligence and drone technology to create detailed spatial maps of refugee settlements in Kakuma and Kalobeyei, Kenya. The innovative project has mapped infrastructure for over 300,000 refugees from more than 20 countries, establishing a replicable methodology for humanitarian organisations worldwide.

Context and Background

The mapping initiative represents a collaboration between UNHCR, the Humanitarian OpenStreetMap Team (HOT), and Microsoft’s AI for Good Lab, with all datasets, models, and code published openly on GitHub for global reuse. Local refugees were trained to operate drones and capture settlement imagery, creating sustainable capacity within the community whilst generating critical infrastructure data.

Machine learning models were used to annotate and identify essential infrastructure including homes, solar panels, and medical clinics across the complex settlement landscape. The project demonstrates how AI can transform urban planning in refugee settlements, enabling more effective resource allocation and sustainable development strategies.

The methodology addresses longstanding challenges in mapping dynamic, rapidly-changing refugee settlements where traditional surveying methods prove inadequate for the scale and complexity of infrastructure needs.

Looking Forward

The open-source approach ensures that humanitarian organisations globally can adapt and implement similar mapping projects without requiring significant technological investment. GitHub’s role in hosting and distributing the tools demonstrates the platform’s potential for connecting developers to humanitarian challenges requiring technical solutions.

The project establishes a framework for using collaborative technology to address complex humanitarian problems, with implications extending beyond refugee settlements to disaster response and urban planning in underserved communities worldwide.

Source Attribution:

Share this article