Advanced Cloud-Native Solution for Efficient Remote Sensing Image Processing using ML/DL Techniques

This project developed a new cloud-native solution to efficiently process the huge volume of available open and commercial Remote Sensing (RS) images using Machine Learning/Deep Learning (ML/DL) techniques in order to achieve state-of-the-art results in various Earth observation (EO) applications. It combines OODA expertise in ML/DL models design and training, satellite images processing and cloud computing fields in order to create a powerful tool able to handle the complexity of satellite images and provide infrastructure to develop new DL capabilities.  The project results and benefits include:

- A new cloud-native solution to efficiently process the huge volume of available open and commercial Remote Sensing (RS) images for ML/DL models training, testing and deployment on distributed infrastructure designed to optimize large models training and ML datasets ingestion and preprocessing.

- An innovative solution to design and develop new ML/DL models and adapt existing RS data preprocessing and processing tools and algorithms to parallel data processing and distributed cloud computing.

- A graphical user interface developed to optimize the visualization of huge volumes of RS images and vector data on a map and manage remote pipeline execution on the cloud while hiding the used distributed infrastructure complexity and heterogeneity to the end user. Our cloud-based RS solution provides access to different processing and storage resources offered by the cloud, can be accessible remotely by many users, is compatible with existing private, public and hybrid cloud architectures and with many satellite missions accessible remotely.