Artificial Intelligence

The Netherlands Plant Eco-phenotyping Centre (NPEC) in Wageningen conducts large-scale plant measurements under controlled conditions to study the interaction between DNA and the environment. This generates over 1,000 terabytes of data. NPEC employs AI to analyse relevant plant components in images while filtering out irrelevant elements. AI is also used for early and accurate disease detection in plants through automated image analysis. Additionally, AI helps identify and explain anomalies in research data, such as detecting and analysing variations in irrigation or genotype effects.

European regulations set limits on AI-powered facial recognition, requiring explicit consumer consent for responsible use. AI can be valuable in food research, for example, by using smartwatches to monitor eating habits more accurately than traditional questionnaires. A step further involves using cameras in care homes to analyse food intake and chewing behaviour, which is only possible with residents' consent. To ensure privacy, AI protocols can be developed to restrict camera footage to the mouth area, avoiding facial recognition while still collecting useful data.

Researchers at Wageningen Environmental Research are improving the World Wildlife Fund's Forest Foresight system, which uses satellite radar and AI to predict where deforestation may occur, up to months in advance. The system detects new roads, which may indicate logging activity, and is being tested in Suriname, Gabon, and Kalimantan. According to university lecturer Johannes Reiche, the system is now also learning to recognise the causes of deforestation, such as mining, agriculture, and logging, while accounting for different forest types. This allows rangers to patrol more effectively and respond quickly to illegal deforestation.

Researchers increasingly use drones and AI for ecological studies. Marine biologist and computer scientist Jeroen Hoekendijk from Wageningen Marine Research applied AI to automatically count seals in aerial images of the Wadden Sea. He is now also using automated image analysis to count seabirds over the North Sea. While initial results are promising, extensive expert data is still needed, particularly to distinguish similar bird species. Hoekendijk bridges the gap between biology and computer science, aiding AI applications such as determining fish age from otolith growth rings—an otherwise labour-intensive process that AI can make more efficient.

- Unfortunately, your cookie settings do not allow videos to be displayed. - check your settings

The Fully Documented Fisheries project is developing a system that automatically records a fishing vessel’s catch without requiring manual input from fishers or observers. The system uses GPS, sensors, an onboard computer, and cameras that capture footage of the catch on the sorting belt. Researchers from Wageningen Marine Research are developing computer vision methods to interpret these images, allowing the system to classify fish species and sizes automatically. In the long run, this technology could enable full, automated documentation of fisheries, contributing to the sustainability of the sector and responsible fish stock management.

Growers have access to increasing amounts of data—on cultivation, greenhouses, sales, labour, and more. However, this data is often scattered across different sources, apps, tools, or external company websites. Wageningen University & Research BU Greenhouse Horticulture is exploring how artificial intelligence (AI) can help growers find connections between these data sources. Researcher Rick van de Zedde explains: “Together, we identify interrelations and how combining different data sources adds value. We bring together experts in IT, cultivation, and energy. They know what to measure, as AI always starts with human expertise.”

The results of Artificial Intelligence can be impressive, but its inner workings often remain incomprehensible. Should we blindly trust these algorithms? Fortunately, no. Innovations in Explainable AI can provide more insight into AI or even enable AI to explain itself.

"Scientists, in particular, want to fully understand their tools. What if AI is biased toward drawing certain conclusions, and that bias initially goes unnoticed?" says Dr. Ir. Bas van der Velden, team leader of Data Science at Wageningen Food Safety Research (WFSR). He and his colleagues are researching ways to make AI more transparent.

- Unfortunately, your cookie settings do not allow videos to be displayed. - check your settings

Read more

• AI doesn’t need to stay a black box - KO magazine 2024