Marvel: Real-time pollen information, FHNW School of Computer Science

Estimates suggest that 10–40% of the world’s population suffer from pollen allergies, and this number is expected to rise as the climate continues to warm. In Switzerland, MeteoSwiss supports allergy sufferers with its real-time pollen forecast.

Our research partner Swisens has made this possible by providing MeteoSwiss with technologies for the detection and identification of pollen particles. These technologies are highly effective at identifying common pollen species in Central Europe. However, extending them to new geographical regions with different plant species remains challenging using current methods.

Swisens aims to deploy its pollen detection system in additional regions around the world. However, as different pollen species occur in these regions, a large volume of data must first be collected. The Marvel project is designed to simplify and accelerate this process.

To develop a pollen detection system that can be deployed worldwide, the Institute of Data Science, FHNW School of Computer Science, is collaborating with Swisens. 

Establishing the current system in Switzerland required extensive manual work to prepare a labelled training data set. This approach is not suitable for global scaling. Our project will therefore develop a machine learning model capable of identifying a new pollen particle, even if it has never encountered it before.

Swisens will be able to integrate this model into an effective pollen forecasting tool and transform it into a market-ready product for people with pollen allergies around the world.

Our Institute’s contribution to the project focuses on machine learning and artificial intelligence. We apply deep learning methods and are extending our approach towards few-shot learning combined with active learning, while continuously collecting and validating increasing volumes of pollen data.

The Marvel project is a strong example of how a university of applied sciences operates: we use state-of-the-art technologies and identify practical applications for them. As a result, our industry partner gains a solution that can be commercialised immediately. People with pollen allergies around the world will soon benefit from even more accurate forecasts and warnings.