Helmholtz AI @ KIDA-KON 2024: AI research in food, agriculture and environment
Helmholtz AI shared its expertise in consulting at KIDA-KON 2024, dedicated to advancing AI applications for food, agriculture, and environmental research. Held on December 2–3, 2024, at the Deutsches Biomasseforschungszentrum (DBFZ) in Leipzig, Germany, this conference is part of the AI and Data Accelerator (KIDA) initiative, which aims to strengthen agri-food research as a core element of the Federal Government's AI strategy.
KIDA – the AI and Data Accelerator in the Field of Food and Agriculture
The KIDA initiative brings together scientists, AI experts, and regulatory stakeholders to address pressing challenges in food and agriculture through cross-institutional collaboration. KIDA-KON 2024 focused on fostering dialogue, sharing innovative research, and providing young scientists with opportunities to present their work and network with experts. This year’s agenda emphasized the transformative potential of AI in achieving sustainable and resilient agri-food systems. Such events play a critical role in addressing societal challenges like food security, sustainable agriculture, and climate resilience through innovative solutions.
Helmholtz AI @ KIDA 2024
In a session that showcased the consulting efforts of KIDA and Helmholtz AI, Tobias Weigel, Head of the Helmholtz AI consultants for Earth and Environment at the German climate computing centre (DKRZ), gave a presentation that illustrated how consulting is provided as an infrastructure. Consultancy services such as the development of ML pipelines tailored against project-specific scientific data, mentoring of research groups and projects, and design and conduct of specialized training workshops were presented for a wide audience. Several examples of consulting work from the research field of Earth & Environment were featured, including the segmentation of tree crowns from aerial imagery, the detection of bowhead whales from hydroacoustic data, and the development of MLops pipelines for model drift detection.
Amongst the current activities of Helmholtz AI consultants in the field is the DeepTrees project at UFZ by Caroline Arnold. The goal of the project is to determine crown sizes of individual trees from aerial imagery of Germany, enabling a wide range of downstream applications. In collaboration with Taimur Khan, consultants refined a pretrained model with manually constructed labels in order to improve segmentation accuracy for the chosen study area of Halle (Saale). The project is still under development with a focus on providing an active learning workflow to better direct additional labelling.
KIDA-KON 2024 highlighted the critical role of AI in addressing societal challenges. Through participating, Helmholtz AI showcased how directed consulting and collaborative efforts can accelerate novel solutions, further emphasizing the importance of AI-driven approaches.