Agro Food Robotics team develops early disease detection system in OPTIMA project

Published on
June 30, 2022

There is a great urgency to reduce the dependence on chemical plan protection products (PPPs) in European agriculture. Such a reduction would decrease PPP use, lower residues, and reduce the impact on human health and the environment. The EU-funded OPTIMA project develops an environmentally friendly IPM framework by providing a holistic integrated approach which includes all critical aspects related to integrated disease management. A Wageningen University & Research (WUR) team of Agro Food Robotics researchers has been in the lead to develop the early detection system for monitoring and localisation of diseases in the field.

“The early detection system uses (spectral) cameras to detect diseases and deep-learning techniques to classify them. In the OPTIMA project we worked with three crops, namely vineyards, apple orchards, and carrots, and focused on the diseases Alternaria and apple scab.”, explains Gerrit Polder, OPTIMA project WUR team lead and Machine Vision & Phenotyping scientist.

“We started collecting images of the three diseases in the field, and we used these images to train the deep-learning algorithm. Once trained, this so-called object-detection algorithm could look at an image and determine whether there is a disease or not, and if there is, where the disease is located.”

A robust real-time smart camera system was developed and tested in field trails in 2021. The system’s detections and GPS location information were sent to a decision support system in the cloud. With the gathered information, a task map could be created and loaded onto a smart sprayer device, which then applied the right dose of plant protection products at the right location, at the right time in the crop.

During the project, the team employed spectral cameras alongside the regular RGB cameras in the system, expecting to get better results in the detection of Alternaria and apple scab. Indeed, the field trials in 2021 showed that, specifically in carrots and apples, adding spectral information significantly improved the detection rate.

Watch the video to learn more about the project and its results from the team members:

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