In sugar beet cultivation, precise yield mapping has not been possible to date. While the current technical equipment of beet harvesters measures the fill level in the machine, this data is not precise enough under all conditions. How much has been harvested where in the field? This is important information, among other things, for the targeted and economical use of fertilizers. The transport chain from the field to the sugar factory could also be organized more resource-efficiently with accurate quantity data.

In the KibEZ project, an AI-based approach for determining sugar beet yield was developed and validated. A training dataset with approx. 16,000 m² at ten locations and a validation dataset with 32 complete bunker loads were created. On this basis, an AI model was developed that determines the beet yield solely from process data of the beet harvester. The results show a correlation between machine data and beet yield and enable yield information without additional sensors. Detailed information in the final report.