Hun Choi ’17
Combining the Human Eye Machine Learning to Better Map African Crop Fields
The Mapping Africa project serves to provide a better understanding of the distribution of crop fields in Africa. The platform currently uses crowdsourcing to classify the crop fields in satellite images, which can be costly in terms of both money and time. The goal of my internship was to develop the proof-of-concept for a crop field mapping platform that combines the current crowdsourcing system and the advanced classification algorithm worked on by former graduate student Stephanie Debats to ultimately classify crop fields automatically and accurately. I spent most of my time figuring out the workflow in Python and QGIS, and used a Maximum Likelihood Classifier to stand in as a rudimentary classification algorithm. I was able to create the entire workflow from taking a satellite image to iteratively improving the classification results. In this way I helped advance active research and lay the foundations for the project’s future. This internship introduced me to the world of research and taught me technical skills in Python, remote sensing, and GIS. Through my work this summer, I confirmed that I wish to continue pursuing computer science, and I also broadened my view of prospective industries I wish to work in in the future.
Technologies for Environmental Study
Civil and Environmental Engineering Department, Princeton University, Princeton, NJ
Lyndon Estes, Associate Research Scholar, Woodrow Wilson School