Sidney Espinosa details how UNHCR, in partnership with GitHub and Microsoft AI for Good Lab, used drone imagery, AI, and open source tools—including GitHub Copilot—to empower refugees in Kenya through intelligent mapping solutions.

Using AI and Open Source to Map Refugee Settlements: The UNHCR and GitHub Story

Refugee camps such as those in Kakuma and Kalobeyei, Kenya, are complex environments lacking critical spatial data, making sustainable development and infrastructure difficult. To address this, the UNHCR (United Nations Refugee Agency), alongside humanitarian organizations and refugees themselves, launched a project to map these settlements using artificial intelligence and open source software.

Turning Drones into Data Maps

  • Local refugees, trained by the Humanitarian OpenStreetMap Team, used drones to collect thousands of high-resolution images across the camp.
  • Initial images were annotated by hand, identifying vital features like homes, solar panels, clinics, and sanitation facilities, creating a reliable “ground truth” dataset.

AI-Driven Feature Recognition

  • Microsoft’s AI for Good Lab developed machine learning models using the annotated data, automating the mapping process across the entire settlement.
  • This drastically reduced the time required to capture and maintain infrastructure information, supporting rapid, data-driven planning.

Open Collaboration on GitHub

  • Every dataset, model, and line of code was openly published on GitHub repositories.
  • The availability of open source materials enables developers and aid organizations globally to adapt, expand, or contribute to the technology for new regions or scenarios.
  • Tools and codebases, organized and cleaned with the help of GitHub Copilot, are now more accessible to a wider audience of civic technologists and humanitarian workers.

Building a Network of Impact

  • GitHub’s collaborative environment connected developers, data scientists, and humanitarian organizations in a global network.
  • This transparent approach ensures knowledge and technical solutions are available for any community facing similar challenges—not just those in Kakuma.

Lasting Results

  • By leveraging the expertise and lived experience of refugees, the technical proficiency of developers, and the organizing power of open source, the project produced tools that help create sustainable, resilient communities.
  • The initiative is a model for humanitarian technology, showing how advanced AI and developer tools, like GitHub Copilot, can make a real-world difference for vulnerable populations.

For more on the project, see Using AI to map hope for refugees with UNHCR, the UN Refugee Agency.

This post appeared first on “The GitHub Blog”. Read the entire article here