14:15 – 14:45
Can artificial intelligence help in automating visual inspection? Yes, it can. Itility and Wageningen UR use Marvin 3D pictures of tomato seedlings and run them through a trained neural network to achieve this. Currently seedling quality is classified by human observers, by observing ten days old tomato seedlings, and defining ‘by eye’ if these will grow into a good or bad tomato plant.
Itility and Wageningen UR automated this manual process. For this, Marvin technology takes pictures of the seedlings from various angles, to create a 3D image. A convolutional neural network is trained to classify these images into good and bad. This trained model is then used to instantly classify seedlings by simply taking a picture.
Marianne Faro is managing director at Itility, and as such leading the analytics competence team of data engineers, data analytics and data scientists within the company. Before joining Itility, Faro has been European information manager at Nike, interim information manager at various municipalities, departmental controller at the automation centre of the Dutch tax office.