IrelandPublic sector2025 – ongoingDepartment of Agriculture, Food and the Marine
Grassland classification for the Area Monitoring System
DAFM's grassland classifier achieved 93% accuracy in an Area Monitoring System study
Problem
Rough upland grasslands reduced the reliability of models detecting grazing and mowing for agricultural compliance monitoring.
Approach
The study trained a support vector machine on parcel-level Sentinel-2 satellite signals and manually labelled grassland data. It separated managed grasslands from rougher land. Integration with the Area Monitoring System improved grazing and mowing detection in the reported evaluation.
Results
Takeaway
Classifying land type first helped activity detection avoid errors associated with rough grasslands. The reported accuracy was a study result, rather than a measure of all compliance decisions.
Categories
- Industry
- Public sector
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and MLComputer vision
- Adoption
- Built in-house
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
Adriaan Keurhorst and Rob O'Loughlin reported the classifier and its integration with activity detection. Their evaluation achieved 93% overall classification accuracy and reported improved grazing and mowing detection.
Full details are in the original sources.
