AI and Sentinell images for agricultural area monitoring - a pilot

Each year, the Swedish Board of Agriculture (SJV) and the county administrative boards make thousands of field visits to pastures with the aim of determining whether the land is cultivated or abandoned. This is very costly and therefore, in the coming period of the Common Agricultural Policy (CAP), the EU proposes that field visits will be replaced by automatic analysis of satellite images.

This project aims to initiate SJV's work to improve quality and efficiency, by developing algorithms and best practices for agriculture monitoring relevant for the management of the CAP. The goal of this project is to develop algorithms that make it possible, in a limited area, to automatically detect whether pastures have been harvested or grazed during the season. By analyzing satellite images using AI / machine learning, changes can be detected and classified to automatically determine the land use over time. In the future, the method can be scaled up to automatically monitor all agricultural land in Sweden.