This repo is forked from the 'Prithvi-EO-Segmentation' code repository developed by Emmanuel Asare of the University of Floria. His repo was developed leveraging the Prithvi Earth Observation (EO) foundational model from NASA ODSI and IBM, specifically for land cover classification in this case. In this case, Prithvi EO was fine-tuned for the detection of artisanal mining activity using Sentinel-2 imagery.
This repository's code is run in four separate steps:
- Sentinel-2 data pre-processing
- Prepration of training data ➡️ this is a direct link to the Google Colab notebook
- Model evaluation
- Land cover mapping