Effective planning and response to humanitarian crises depend on the knowledge of the local situation and the spatial distribution of populations. However, fine-scale population maps are often unavailable and not easy to generate. Especially in contexts affected by violence and conflict, maps and census information are often incomplete or outdated.

The project aims to estimate local size and density, as well as related information, such as settlement type and population changes, from satellite images and social media posts and integrate them into detailed, spatially resolved population maps. More information can be found in this paper.

ETHZ PI: Prof. Konrad Schindler, Photogrammetry and Remote Sensing, D-​BAUG;
EPFL PI: Prof. Devis Tuia, Environmental Computational Science and Earth Observation Laboratory, EPFL

Partners: Thao Ton That Whelan, ICRC and Ferda Ofli, Qatar Computing Research Institute

Photo: ICRC

• Preliminary results of the proposed methodology were presented on August 15th of 2021 at the ACM SIGKDD Conference on Knowledge Discovery and Data Mining, and on May 27th of 2022 at the Living Planet Symposium.

• The paper presenting the proposed methodology for population estimation was published, on 22 November 2022 in the Scientific Reports of Nature Journal

• Swissinfo

• 20 minutes

• The project results were also presented at the Science Diplomacy week