Leveraging AI technology and utilising Earth observation data to extract valuable insights is challenging. The need for high-performance cloud environments for model training and inference, the scarcity of suitable and accessible Training Datasets used to train AI models to perform specific tasks, where the main barrier is that gathering and labelling EO data is a convoluted process, or the need for libraries and environments that allow streamlining the training of models specifically designed to use EO data are just some of the problems that AI4EO engineers face.

With the aim of alleviating the pain points of this entire process and encouraging the development of applications that extract valuable information from Earth Observation data through AI, trying to generate a positive impact, EarthPulse has developed a set of open source solutions aimed at democratizing AI4EO, covering everything from the generation and labeling of a Training Dataset to the training of a model. We'll dive into SCANEO, an intelligent and AI-powered standalone labeling tool for spatial data; EOTDL, a complete environment funded by ESA that allows the creation, exploration, download and upload of both Training Datasets and pre-trained ML models for EO applications, and PyTorch EO, a Python library that aims to make Deep Learning for Earth Observation data easy and accessible to real-world cases and research alike.