En standardiserad inventeringsmetod för att uppskatta skador på gröda: med fokus på vilt – och vildsvinsskador

Current damage assessments are time-consuming and costly, often carried out on foot with the risk of subjective results. SLU (Swedish University of Agricultural Sciences) has been tasked by the Swedish Environmental Protection Agency (as of March 23, 2022) to develop a standardized method for monitoring wildlife damage in agriculture. The developed method is based on drones equipped with cameras and sensors for visible and invisible light, capturing high-resolution 2D or 3D images of fields, which are then analyzed using AI to identify, classify, and measure damage. The goal for the method is to be applicable at a local farm level but also at large areas such as wildlife management units (ÄFO) or larger, which may become important for decision-making in wildlife and agricultural management. The AI-interpreted images serve as the basis for a report containing information about the origin, extent, and cost of the damage, the results of which can be used both locally by farmers and nationally in a future monitoring system. The system is currently semi-automated but can be further optimized with high-performance computers. So far, the project has shown promising results, particularly in identifying wild boar damage, which is the most common damage type in our image database collected, so far. However, for the method to become fully applicable, the database needs to be complemented with more images of other damage types (drought, water, machine tracks, etc.) and for other wildlife species and crops is required than what has been gathered during the limited project time. The remaining technical challenges therefore involve automating all steps from image analysis to reporting in a more powerful computing environment, as well as training the algorithm with more images of other types of damage besides wild boar. Finally, purely organizational issues regarding a potential future wildlife damage assessment platform for crop damage — its governance, legal possibilities, limitations, and costs, are all outside SLU’s mandate, Given that they all can be solved, it is SLU’s evaluation of the developed assessment method that it represents a fully realistic way forward, with potential to be further developed for use with airplanes and, in the future, perhaps satellite imagery. The method could thus become an important future fact-based wildlife management tool, benefiting individual farmers, hunting rights holders, wildlife managers and authorities.