4. juli 2023
Forklarlig AI skal disrupte kornindustrien og sikre tillid blandt landbrugere
Landbrugssektoren er den mindst digitaliserede sektor i verden, og en stor del af kvalitetssikringen af fødevarer foregår stadig manuelt. Et forskningsprojekt skal styrke forståelsen for og tilliden til AI og billedanalyse, som kan forbedre kvalitetssikringen, fødevarekvaliteten og optimere produktionen.
En af de helt store kritiske barrierer ved at bruge AI og billedanalyse i landbrugs- og fødevareindustrien, er tilliden til, at det virker.
I dag er den manuelle visuelle inspektion af korn stadig en af de vigtigste kvalitetssikringsproducerer i hele værdikæden for at bringe korn fra marken til bordet – og sikre, at landbrugeren får den rigtige pris for sine korn.
Can you find ‘Okapi’ in these pictures? Ph.D. student Lenka Tetková from DTU uses this example to explain how image classification works.
An important competitive advantage
As a global producer of niche products, FOSS must always stay two steps ahead of competitors.
– To ensure there is a market for us in the future, it is crucial to be the first with new solutions. It is challenging to make a profit if there is already a player doing it better, which is why we constantly introduce new digital technologies to improve our analysis tools. And here, collaboration with researchers from the country’s universities is very valuable to us, as we gain new insights and proposed solutions for the further development of our tools, says Erik Schou Dreier and continues:
– In this project, we hope that collaboration with researchers will lead to the development of AI methods and tools that enable us to create new solutions for automated image-based quality assessment and, secondly, that we can increase trust in our product with explainable AI. It is one of the critical themes for us—to create a product that is trusted.
Facts about FOSS
FOSS’ measuring instruments are used everywhere in the agriculture and food industry to quality assure a wide range of raw materials and finished food products.
Traditionally, light wavelengths are measured, and the measurements are used to obtain chemical information about a product. This can include knowledge about protein and moisture content in grains or fat and protein in milk, etc.
FOSS’ customers are large global companies that use FOSS’ products to quality assure and optimize their production—and to ensure the right pricing, so, for example, the farmer gets the right price for their grain.
Deep Learning and Automation of Imaging-based Quality of Seeds and Grains
Project Period: 2020-2024
Budget: DKK 3.91 million
Project participants:
Lenka Tetková
Lars Kai Hansen, Professor DTU
Kim Steenstrup Pedersen, Professor, KU
Thomas Nikolajsen, Head of Front-end Innovation, FOSS
Toke Lund-Hansen, Head of Spectroscopy Team, FOSS
Erik Schou Dreier, Senior Scientist, FOSS
What is a Deep Learning Neural Network?
Deep learning neural networks are computer systems inspired by how our brains function. It consists of artificial neurons called nodes organized in layers. Each node takes in information, processes it, and passes it on to the next layer. This helps the network understand data and make predictions. By training the network with examples and adjusting the connections between nodes, it learns to make accurate predictions on new data. Deep learning neural networks are used for tasks such as image recognition, language understanding, and problem-solving.
