Every spring, millions of people brace for the first sneeze of allergy season, while farmers prepare for the threat of fungal disease. What if a single technology could protect human health and farm yields at once?
Airborne particles like pollen and fungal spores may be microscopic, but their impact is vast. Pollen triggers allergies that affect hundreds of millions of people worldwide, while spores spread plant diseases that destroy up to 30% of global crops annually, worth billions. Conventional monitoring systems rely on traditional, manual lab analysis: samples must be collected, transported, and examined under a microscope, often taking days. The result? Delayed forecasts, missed prevention windows, and excessive use of chemicals in agriculture.
To tackle this, CSEM recently teamed up with Swisens through the European-funded AGRARSENSE project, working with Agroscope and the University of Applied Sciences and Arts of Northwestern Switzerland (FHNW). Swisens had developed its SwisensPoleno platform, an advanced system that captures multiple data points from each pollen or spore particle: a holographic “3D-like” snapshot from two sides, fluorescence spectra (the range of light the particle emits), fluorescence lifetime (how long the light lasts), and depolarization (how the particle alters the light’s orientation). These measurements create a detailed fingerprint to distinguish specific spore and pollen types.
Using this data, Swisens and CSEM built software for automatic classification, where patterns in the measurements label each particle type without manual sorting. CSEM’s engineers developed a multimodal deep learning model, the AI “brain,” which merges the holographic snapshots and light measurements into a richer representation. “Seeing both clues at once is the real step forward, because it finally allows different spore types that once looked almost identical to be separated,” explains Silas Dietler, R&D Engineer, CSEM.
In parallel, CSEM explored a more experimental approach, where AI discovers patterns in raw data with far less human input. “This self-learning approach is a promising direction for future systems,” adds David Hemmi, Head of Research & Business Development, CSEM, “but today the focus is on the multimodal model we developed together with Swisens.”
The self-learning approach uses real measurements in the field so that the AI can adapt as conditions change with local plants and weather. New pollen or spore types can be added quickly without rebuilding the model, making future rollouts more practical and scalable. The system thus keeps improving while performing consistently for known pollen or spore types.
The implications reach far beyond the lab. For allergy sufferers, more accurate pollen detection means better and faster forecasts, empowering healthcare providers and individuals alike. For farmers, real-time spore monitoring enables antifungal treatments that are better targeted, reducing chemical use while maintaining yields.
The benefits for society and the environment are clear: fewer serious breathing-related health problems, lower emissions from farming, and more robust food systems that can better withstand shocks. “This technology helps us bridge the gap between environmental health and human health,” adds Erny Niederberger, Head of Science, Swisens. “Working with CSEM allowed us to take a major step forward, from a local innovation to a globally scalable solution.”
CSEM’s multimodal AI model pipeline underpins the system now in use, automatically identifying particle types from SwisensPoleno measurements. By combining AI with Swiss-engineered hardware, CSEM and its partners are delivering technology that protects both people and the planet. As Erny Niederberger concludes, “Our goal isn’t just to build smarter sensors; it’s to help societies act faster, more intelligently, and more sustainably.”
Working with CSEM has significantly advanced our work. Their AI model brought additional flexibility to our particle classification, enabling real-time classification of bioaerosols smaller than pollen. This collaboration is helping make agriculture truly data-driven.
Explore CSEM’s data and AI expertise to discover how embedded AI, edge computing, and secure data pipelines transform raw data into actionable intelligence. Or contact us through our form to discuss your digital transformation.