Skoči na vsebino

Crop Growth Monitoring and Analytics Services

Opis

Relying on Copernicus data, fused with weather predictions and soil parameters, this service provides comprehensive monitoring and analytics of crop development in agricultural fields. By utilizing a robust normalized difference vegetation index (NDVI) forecasting model, with embedded data cleaning and cloud removal mechanisms, it provides users with a time series of crop development predictions for any given land parcel the user is querying about, together with anomaly detection on past crop behaviour for pests and disease detection and forecasts of the expected crop yield. The service combines pipelines for time series preprocessing, forecasting, and continuous vegetation monitoring. A cloud removal and gap-filling pipeline leverages a Masked AutoEncoder (MAE) transformer architecture with temporal embeddings to reconstruct obstructed historical satellite observations and is integrated into a forecasting pipeline that employs ConvLSTM and ConvGRU architectures with temporal skip connections to project NDVI trends forward in time. A behaviour change pipeline synthesizes these outputs to detect anomalies and estimate expected crop yields.

Pričakovani rezultati

When querying the service for a particular land parcel, the retrieved results are given in the form of a time series of past and predicted NDVI-based crop development indices, with marked possible anomalies in past behaviour, expected reasons for them, and the overall prediction of crop yield.

Kartica storitve

KategorijaAI products
DomenaPrecision Agriculture, Crop Monitoring
VertikalaGreen Transition
Ciljni uporabnikiPrecision Farming Solution Providers (SMEs), Advisors, Farming Associations
Tip storitveService
RezultatTime series of past and predicted NDVI-based crop development indices; anomaly detection reports; crop yield predictions per land parcel.
Zahteve
Omejitve
Na voljo odQ3/2027
Licenca
Slikaassets/service-visuals/crop-growth-monitoring-and-analytics-services.png
Oznaka slike