NPEC, or the Netherlands Plant Eco-phenotyping Centre, stands as a beacon of innovation in plant research. Focused on generating valuable insights through data-driven experiments, NPEC employs cutting-edge technologies to address the challenges faced in agriculture and plant research. Rick van de Zedde sheds light on the intricate layers of AI implementation within NPEC.
“Imagine one of those Matryoshka nesting dolls. You lift the lid on one and there is a whole new layer to explore inside of it. And so on,” says Van de Zedde, Chief Technology Officer NPEC and Machine Learning and AI expert for the Vision + Robotics programme at Wageningen University & Research (WUR). “We started by generating a vast amount of data through our experiments. The initial assumption was that more data equaled more value. But we quickly learned that researchers are more interested in answers than raw data.”
Image-based AI detects features of tomato plants
In a second layer, all that raw data can be turned into information using image-based AI. Van de Zedde explains, “We use AI to dissect the data further. For example, in our greenhouse experiments with tomato crops, we can precisely measure aspects like leaf surface area or the architectural features of a plant. We teach the AI to recognise different components – leaves, stems, pots, and sticks – allowing us to turn the heaps of data into organised and insightful information.”
The third layer involves ensuring that the experts in WUR’s Vision + Robotics programme have the relevant AI knowledge. These experts play a crucial role in translating raw data into meaningful insights. AI helps identify and categorise elements within images, providing researchers with targeted and relevant information.
Acknowledging the challenge of dependence on specific AI experts, Van de Zedde unveils the fourth layer. “To overcome this, we have invested in training initiatives. Our goal is to share our expertise, ensuring a broader group of researchers and R&D staff in businesses possess the relevant AI skills. This reduces reliance on a single super expert and enhances our collective impact.”
User-friendly tooling for AI in plant research
So what’s the next layer in NPEC’s AI Matryoshka doll? Looking ahead, Van de Zedde envisions the fifth layer as a comprehensive framework. “We’re working towards a user-friendly system where computers facilitate seamless data transfer and users can easily train AI models. The aim is to make AI accessible to a broader research audience. We want to empower users to derive valuable insights without deep technical AI expertise, with correct use promoted by the developed tools.”
In essence, the layered approach at NPEC ensures that AI becomes an integral part of the research process, a framework to be built upon by researchers to come. By progressing from raw data to meaningful insights, NPEC’s journey is one of inclusivity and sustainability in the realm of high-tech innovation in life sciences.
NPEC is at the forefront of plant research, leveraging ultra modern facilities to conduct experiments in controlled environments like greenhouses and climate cells, as well as in open fields with drones, for example. The Centre’s primary goal is to provide high-level research with meaningful data and insights. Those insights then contribute to the advancement and acceleration of plant research, benefiting not only Wageningen researchers but also the broader scientific community. Read more on the NPEC website.
Photo header: Guy Ackermans