BaKIM wins first place in open-source competition for AI
BaKIM, the open-source forest monitoring app of the City of Bamberg and GFA’s Digital Innovation Unit (DIU), won first place in the new AI category of the Open Source Business Alliance’s Open Source Competition at SCCON 2026. It turns standard drone images into tree-level data.
BaKIM replaces largely analog forest inventories with an automated pipeline: Drone-based RGB aerial images are analysed by three computer vision models that detect individual trees and determine their vitality and genus. The results come as open GeoPackage files that work in any Geographic Information System (GIS). This frees up time for foresters to care for their stands and gives municipal councils a data basis for decisions on staffing needs in forest and tree care.
The City of Bamberg trained the models; GFA’s DIU team brought them into production and integrated them into a web application for end users. The application is now in productive use by 19 institutions and municipalities. In the city of Bamberg its core system has been part of the forestry office’s processes for three years.
BaKIM builds on existing open-source repositories and open-weights models, runs on scalable GPU resources of a European provider, and is published on openCode. The BAMForest training dataset is part of the global MillionTrees project and already used in Frankfurt and Denmark. Next steps are models for biomass estimation and further forest structure parameters.
It was impressive to see that not only the application, but also the models and training data have been made available as open source and integrated into global research projects, concluded the Jury of the Open Source Business Alliance.
> Find out more on the BaKIM project on our website.
