Swisens is the world leading instrument manufacturer in the field of bioaerosol monitoring. They have equipped the Swiss monitoring network with about 20 instruments (e.g., Poleno) and they are now doing the same in Ireland and other counties worldwide. The Poleno monitor relies on holography and machine learning to identify pollen and fungal spore particles. The morphology of these particles differs from country to country and the Poleno algorithms need to be re-trained in each country with labour-intensive and extensive field campaigns. The reference method with which the Poleno is compared is the Hirst impactor, a method known for its high uncertainties (>30%). The idea behind this proposal is to: 1) implement new algorithms (e.g. zero-shot learning techniques, few-shot and semi-supervised learning techniques) so that the monitoring system can be deployed in new regions with new bioaerosol types without expensive data labeling efforts; and 2) establish a new reference method for particle counting in the size range relevant for pollen (>15 µm) for calibrating the Poleno in the laboratory.
