HOLSTEIN (Holistic approach to the sustainable provision of livestock health in Lower Austria) is a project funded by the State of Lower Austria. It pursues a holistic strategy for securing veterinary care in rural areas, combining curative medicine with preventive herd monitoring. The work spans evidence-based telemedicine and teletriage infrastructure, sensor networks for early disease detection, and digital tools that make food-animal practice more attractive and sustainable, ultimately improving animal welfare and resource efficiency.

PetScan establishes a smartphone application for preliminary veterinary diagnostics, developed to support veterinarians and empower pet owners with early disease detection and is funded by Kärntner Wirtschaftsförderungsagentur. It combines optical AI with complementary sensor-based analysis (XPS) to provide an initial assessment from an image, indicating whether professional ttention is advisable. By lowering the technical and cost barriers to first-line screening, PetScan reflects the group's focus on applicability – putting complex diagnostic technology into the hands of non-expert users.

ROBOKIZ (Robust automatic analysis of drone images in plant breeding using artificial intelligence) is funded within the FTI Strategy Lower Austria 2027. It develops AI methods for analyzing high-resolution drone imagery in crop breeding, including a high-precision imaging pipeline for wheat phenotyping that captures oblique side-view images of wheat heads to detect Fusarium Head Blight. A metadata-driven cleaning process reduces the captured data to a curated set, enabling scalable, reproducible, image-based assessment that replaces labor-intensive manual scoring.

SalaCount addresses individual identification of wild salamanders for biodiversity monitoring, where limited annotated data and challenging field conditions make conventional methods difficult. It introduces a model-free re-identification approach based on each animal's unique color patterns, extracted through blob detection and Hu Moments rather than trained deep networks. By avoiding the need for large labeled datasets, the method supports robust, low-cost wildlife tracking and demonstrates how computer vision can advance ecological research and species monitoring.