---
title: README
emoji: 🚀
colorFrom: indigo
colorTo: red
sdk: static
pinned: false
---



The PhilEO project aims to develop geospatial AI Foundation Models for Earth Observation. Foundation Models are a paradigm shift from task-specific AI models. They are trained in a self-supervised way on large scale data. Then they can be adapted to various downstream tasks by fine-tuning, hopefully with few samples and performances equal or better than fully supervised AI models.


The first asset of the PhilEO project is the PhilEO Bench, a benchmark of geospatial tasks to evaluate the performances of State-Of-The-Art Foundation Models on diverse downstream applications.

PhilEO Bench: Evaluating Geo-Spatial Foundation Models - [PhilEO-Paper](http://arxiv.org/pdf/2401.04464.pdf)


Powered by European Space Agency (ESA), Phi-lab: [ESA-Philab](http://philab.esa.int/)