The First Version of the STELAR Toolkit Now Available!
STELAR, a 3-year Horizon Europe project focused on innovative data management approaches in agriculture, made available the first version of its state-of-the-art toolkit, which supports functionalities such as extracting structured information from food safety reports and fusing satellite and field sensor data to improve crop classification, yield prediction and suitability map construction. The project itself is also developing workflows for the same purpose, as well as a platform for publishing and discovering metadata about datasets in the agrifood sector, as well as linking datasets with data processing workflows for Machine Learning and AI applications.
STELAR’s consortium, gathering experts from Greece, Germany, Netherlands, Italy, and Serbia, led by Athena Research Center, has invited all data scientists and analysts, as well as data aficionados, to explore the tools developed so far!
STELAR Toolkit: Smart Data Tools Available to All
Those interested in the first version of the STELAR Toolkit can inspect the included tools.
There are currently more than 10 tools available on STELAR’s GitHub repository, some of which are:
- Data Profiler – A software component that automatically extracts metadata from various dataset types and outputs it in HTML, JSON, and RDF formats to enhance data description and facilitate discovery and exploration.
- Synopses Data Engine – A tool for computing data summaries over high-speed data streams.
- KGSEC – A software component that takes a Knowledge Graph (KG) as input and produces, as output, an implicit schema of the graph that can facilitate exploration, query formulation and insight extraction.
- pyJedAI – A python-based tool for data integration that supports numerous established and new techniques for schema and entity matching, supporting both textual and geospatial data.
- TS-Impute – A time series imputation tool that aims to improve the quality of time series data, e.g., for filling gaps in Earth Observation time series that are caused by the presence of clouds.
- EO Data Fusion – A tool for the alignment and fusion of Earth Observation data.
- Correlation Detective – A fast and scalable family of algorithms for finding interesting multivariate correlations in large datasets.
- Food Entity Extraction tool – A tool that performs named entity recognition and text classification leveraging a large variety of techniques, from traditional string matching to Large Language Models.
- Field Segmentation tool – A tool that is able to segment satellite images into individual fields.
- Crop Classification tool – A tool for labelling satellite images into different crop types.
The STELAR Toolkit will be integrated into the STELAR KLMS (Knowledge Lake Management System), while the tools can also be used as stand-alone components. The Toolkit has an open-source code that is available on GitHub.
The STELAR Project: Connecting the Dots in Food and Farming
The STELAR project is developing a Knowledge Lake Management System (KLMS) fit for today’s agrifood data needs. Apart from also developing tools and workflows for this purpose, it’s testing its progress and solutions through three real-life Pilots.
The first Pilot is led by the Greek data and analytics company (Agroknow) and focuses on data-driven food safety in supply lines, using AI to predict risks and prevent food recalls. The second Pilot is headed by VISTA, an innovative German company in the field of remote sensing and aims at early crop growth predictions using Earth Observation data. The third Pilot is overseen by the Italian company ABACO, a leader in development of software solutions for the management and control of land resources, and it targets data-driven decision-making in precision farming.
"For the seamless progress of the Pilots, STELAR is taking a user-centred approach focusing, among others, on customer satisfaction, software quality, and environmental impact. Overall, STELAR aims to design, develop and evaluate a Knowledge Lake Management System (KLMS) to empower users with data that’s easy to find, access, and reuse, as well as data that is interoperable, high quality, and reliably labelled."
Dr Dimitris Skoutas, STELAR’s Project Coordinator and Principal Researcher at Athena Research Center
STELAR’s Road Ahead: More to Come
STELAR, co-funded by the European Union, will run until August 2025. The consortium consists of Athena Research Center, National and Kapodistrian University of Athens, Eindhoven University of Technology, University of the Bundeswehr Munich, Altair RapidMiner, Agroknow, VISTA, Abaco and Foodscale Hub.
Make sure to stay tuned for more updates on STELAR use cases and the project in general! Follow closely the project’s website, as well as its LinkedIn, Facebook, Twitter, and Instagram pages.