Artificial Intelligence (AI) and Machine Learning (ML) algorithms have great potential to advance processing & analysis of Earth Observation (EO) data. Training datasets (TDS) are crucial for ML and AI applications but they are becoming a major bottleneck in more widespread and systematic application of AI/ML in EO.
The issues include:
Another obstacle to the use of AI/ML in EO applications for non-EO experts is a lack of domain-specific knowledge such as map projections, file formats, calibration and quality assurance. As such, AI-Ready EO Training Datasets (AIREO) should be self-explanatory, follow FAIR principles and be directly ingestible for AI/ML applications.
Our approach is to review current initiatives, activities, techniques, tools, practices and requirements for preparing, using and sharing AI-Ready EO TrainingDatasets. We have set up the AIREO network of stakeholders and practitioners in the AI/ML, EO and data science communities and from other relevant science disciplines. We have captured community requirements through an online survey and expert consultations. And we have developed a number of resources for the AI/EO communities, including specifications, best practice guidelines, a python library for ingesting TDS into workflows and pilot datasets and accompanying Jupyter notebooks.
The AIREO specifications, best practices and datasets are designed to meet FAIR (Findable, Accessible, Interoperable, Reusable) data principles and to involve and build on top of relevant community initiatives.
The AIREO activity is sponsored by the European Space Agency, and the partners are the Irish Centre for High-End Computing and the Centre for Applied Data Analytics and Artificial Intelligence.
ICHEC is Ireland’s national centre for compute and data management platform and services, and is legally hosted by NUI Galway, a public university. ICHEC hosts and operates the national supercomputer (Kay) for industrial and academic users. In addition, ICHEC provides access through SPÉir (Satellite Platform for Ireland) to Sentinel-1,2,3,5p, Aeolus and Envisat including historical data with unlimited bandwidth and parallel downloads. The platform also offers data access, HPC and AI-aided services around the EO datasets.
ICHEC has expertise in EO, AI and HPC. Remote Sensing of Irish Surface Waters (INFER) and Remote Sensing of Macrophytes (MACRO-MAN) are two Environmental Protection Agency (EPA) projects that involve the use of GIS and EO data mainly focused on European Space Agency (ESA)’s Sentinel-2 data. ICHEC is developing the SPÉir online platform to allow national users to access the Sentinel data archive and runs the ESA Validation Data Centre (EVDC) which hosts EO data sets for atmospheric applications. Other projects include the application of radar (Sentinel 1) to map flood events, invasive species mapping using optical data (Sentinel-2) and detection of events using edge-processing onboard micro and nanosats.Visit website
CeADAR is Ireland’s National Centre for Applied Data Analytics and Artificial Intelligence. It is a market-driven technology centre for the development, and deployment of data analytics and AI technology and innovation. The primary outputs of the Centre are products, prototypes, demonstrators, technology reviews, collaborative research projects and bespoke applied projects. CeADAR has particular strengths in predictive analytics, optimization, machine and deep learning, real time analytics, visualisation, blockchain and smart contracts. CeADAR produces applied research projects at high Technology Readiness Level (TRL) and actively participates in promoting AI in Ireland and in developing consortia at EU level. CeADAR is the designated EU AI Digital Innovation Hub in Ireland ( https://ai-dih-network.eu/) and is one of only 30 across the EU, as well as holding a European i-Spaces GOLD award from the Big Data Value Association ( https://www.bdva.eu/i-Spaces).
CeADAR have particular expertise in the application of AI and machine learning to Earth Observation data, having worked on a range of applied machine learning projects using EO data with it’s industry partners, as well as ‘open innovation’ demonstrator projects such as a machine learning pipeline for integrating EO imagery with complimentary annotation data sources, and the automation of state-of-the-art best practices for training and evaluating deep learning models for EO imagery (https://www.ceadar.ie/pages/ai-for-eo).Visit website
More on AIREO Resources (Specifications, Best Practice Guidelines, Pilot datasets and Jupyter Notebooks and Python Library)
Learn more on our network and how to get involved in AIREO activity visiting the Community activity page
To subscribe to the AIREO network or to contact the AIREO Team please email email@example.com