AgroVeFe in English

// NRRP research project · FAZOS · 2025–2029

From satellite images to maize yield potential maps

AgroVeFe develops machine learning models that estimate within-field maize yield potential from free Sentinel-2 satellite imagery. The results are zone maps that agricultural machinery can load directly for variable-rate seeding and fertilisation – without expensive yield sensors.

11

scientific papers in the first year

56

combinations of vegetation indices and phenological metrics

6

seasons of yield data (2020–2025)

R² ≥ 0.7

target prediction accuracy

How it works

01

Satellite

Sentinel-2 time series from the Copernicus programme, cloud-masked and harmonised to 20 m.

02

Indices and phenology

Eight vegetation indices, including saturation-resistant ones, and seven phenological metrics of crop development.

03

Machine learning

Random Forest, SVR, Cubist and ensemble models, trained on precision-harvest data and rigorously validated.

04

Map for the machine

Yield potential zones in ISOXML format, complemented by soil electrical conductivity scanning with RTK GNSS.

Latest news

Partners

The project is carried out by the Faculty of Agrobiotechnical Sciences Osijek (Josip Juraj Strossmayer University of Osijek, Croatia), in scientific cooperation with the Faculty of Agriculture, University of Belgrade (Serbia), and the Agricultural Institute of Republic of Srpska, Banja Luka (Bosnia and Herzegovina).

We are also looking for farms willing to share yield data or provide a field for testing – find out more. Contact: iplascak@fazos.hr

AgroVeFe (581-UNIOS-30) is funded by the European Union – NextGenerationEU through the National Recovery and Resilience Plan 2021–2026. Views and opinions expressed are those of the authors only and do not necessarily reflect those of the European Union or the European Commission. Neither the European Union nor the European Commission can be held responsible for them.