AI4Agri Knowledge Graph
A dataset from Earth Observation, Machine Learning models and vineyard data, in the form of a Knowledge Graph.
This data set has been produced for the AI4Agricultural pilot in the AI4EU project, and contains all the data used and produced by that pilot. This pilot is targeting a specific application in the scope of precision agriculture to help predicting the yield and assessing the quality of the production in vineyards using remote sensing data and AI models. We have designed an OWL ontology to model the resources and then interlinked the different data both in the temporal and spatial dimensions.
The dataset contains in the form of a Knowledge Graph information coming from Earth Observation, Machine Learning models and vineyard data, that were used in the AI4Agriculture pilot. The dataset is distributed in RDF N-Triples format. We also provide the OWL ontology that was used to model the data.
The Knowledge Graph (KG) contains information from various sources regarding different attributes of grapes in specific parcels of the sample vineyards. Based on the geospatial information of each sample area, we were able to link information coming from Earth Observation, Machine Learning models and vineyard data.
The KG contains the following data for the years 2020 and 2021, that were used in the pilot:
- Vineyards data (parcels, sample areas, petiol analysis, production, soil metadata, grape maturity)
- NDVI (corrected NDVI values based on Satellite images)
- Drone Images (geolocated plant images from the sampling areas)
- Number of Clusters (clusters of grapes detected in the plants of the sample areas, based on a counting model and drone images)
- Yield and Quality (data for each sampling area calculated by a model that uses the above sources along with meteorological data)
The classes and properties that were created to model and link the data are available in the OWL ontology that is provided along with the KG (RDF N-Triples format).