The PRONTO (Predicting Potato Sprouting to Optimise Tuber Storage) project, already introduced in a previous news article, addresses one of the key challenges currently facing potato storage.
Following the 2020 ban on chlorpropham, a chemical previously used to prevent premature sprouting in stored potatoes, the sector has had to identify new strategies to preserve product quality and extend storage life. One of the most widely adopted alternatives is orange oil, a natural compound extracted from orange peels. While effective and safe for consumers, it is relatively costly and highly dependent on timing: applied too early, its effect may not last long enough; applied too late, it may not be effective.
To address this challenge, the Innosuisse-funded PRONTO project has developed machine learning models capable of detecting the early signs of sprouting and supporting more accurate management of anti-sprouting treatments.
Over the past few months, the research team has focused on improving the models’ performance and their ability to adapt to real-world conditions. “While the models perform very well under certain conditions, they can face challenges when factors such as seasons or cultivation batches change,” explain Omran Ayoub and Davide Andreoletti, researchers at the Institute of Information Systems and Networking (ISIN).
To overcome this limitation, the team worked on improving the models’ ability to generalise across different scenarios, adapting established machine learning techniques to the agricultural context. This has made the models more robust and capable of maintaining reliable performance even under conditions that differ from those encountered during development.
The results have enabled the integration of the developed solutions into the systems of industrial partner Vivent Biosignals, a Swiss agritech company specialising in technologies for monitoring plant health and status. The models are already being used in operational environments and are currently undergoing field evaluation, a key step in assessing their effectiveness under real farming conditions.
“The project is a concrete example of technology transfer, where research is translated into practical tools. Our collaboration with Vivent Biosignals has allowed us to bring research outcomes into a real-world setting, contributing to the development of innovative solutions for more sustainable agriculture,” the researchers conclude.