DigiFarm is a Norway-based agricultural technology start-up that wants to help farmers take better decisions, increase their yield, and reduce costs all at the same time. The team develops machine learning models that provide agricultural B2B partners with real-time data insights that are designed to improve their operations. Through the use of simple and dependable digital tools, farmers benefit from reduced seasonal uncertainty, minimised production costs, and increased crop yields. DigiFarm has developed a state-of-the-art field boundary detection model based on deep neural networks and a super-high resolution (0.25m) data set that makes their model 40 times more accurate than existing Sentinel-2-based models. This is how the company provides highly accurate in-field data to clients. Moreover, it substantially increases the accuracy of field boundaries and planted acreages. The high level of training data available from Norway serves as the basis of the model. The DigiFarm team has now tested the system and achieved the same high level of accuracy in the US, France, and Brazil. The next product in DigiFarm’s pipeline is a crop classification and zoning model that will enable farmers to implement variable seeding rates, spraying, and yield predictions.

Nils Helset, Konstantin Varik