Description

In this project, I utilized Landsat-8 satellite imagery and demographic data from the Demographic and Health Surveys (DHS) to predict asset ownership levels across Africa. I trained a ResNet-18 model, leveraging transfer learning to adapt the architecture for a regression task. My model outperformed traditional linear regression models, though it showed signs of overfitting. Despite this, the predictions provided valuable insights into regional asset ownership patterns. This project highlights the potential of combining satellite imagery and machine learning to understand socioeconomic conditions in developing regions.

Poster presentation

Github Repo