Use case · Pilot
Making research data ready for AI
Research, experiment and robotics datasets become accessible for training and benchmarking AI, connecting CEADS with AgrifoodTEF, the Agriculture of Data partnership and EOSC.
Challenge
Valuable datasets from experiments, test farms, robotics and AI trials stay in project silos. SMEs that want to validate an AI solution against sustainability criteria cannot find comparable training and benchmark data, and research infrastructures each solve semantic interoperability on their own.
Actors
Led by INRAE with Fraunhofer IESE, FBK, Universitat de Lleida, ILVO, PSNC, ANAMOB and Josephinum Research. Research infrastructures and testbeds in France, Italy, Spain, Belgium and Poland, agritech SMEs and AI developers.
Data
Experimental, sensor, imagery and robotics datasets from research infrastructures, annotated and harmonised for algorithm training and benchmarking, shared in controlled environments.
How CEADS enables it
CEADS links testbeds and research infrastructures across the EU and makes their datasets discoverable with clear provenance, quality and permissions. AI-assisted image segmentation and annotation, with expert review, turn agricultural imagery into structured, semantically annotated datasets; metadata enrichment and harmonisation tools reduce manual preparation.
Value created
Underused research data train and benchmark better AI and robotics, technologies such as semantic interoperability are shared with other programmes, SMEs can validate solutions against CAP-aligned sustainability criteria, and AI adoption reduces labour burden and creates new services.