News

Camera on tractor sees whether biological crop protection is necessary

Published on
December 15, 2021

In the case of diseases and pests in crops, it is important to intervene as quickly as possible. And it would be even better is action could be taken immediately after detection. The Business Unit Greenhouse Horticulture of Wageningen University & Research is working with a number of European partners on the development of fully automatic biological crop protection to this end. WUR is looking at self-learning computer models that decide whether biological crop protection is necessary on the basis of digital images.

Within the European OPTIMA project, WUR has been working together with partners from Greece, France, Italy, Spain, Belgium and Portugal since 2018. Each partner focuses on a different part of the research project, such as the spraying technique or a decision support system. The research focuses on downy mildew (in grape), scab (in apple) and Alternaria (in carrot). WUR investigates if it is possible to recognize diseases at the earliest possible stage, preferably even a few hours after a disease occurs. The aim is for the cameras to be mounted on a tractor for example. While driving through the crop, the camera recognises whether there is an infection. Subsequently, biological crop protection is applied in conjunction with the decision support system.

WUR developed three so-called 'smart cameras' for the crops. These cameras can recognize diseases - through image recognition and algorithms. To this end, cultivation experts assessed photos of the crops in question: they indicated the position of infections in the photos. With deep learning algorithms, the cameras learned to recognise diseases. Last year, three RGB cameras and a multi-spectral camera were tested, among others at the European partners and several companies. The multi-spectral camera visualises eight different colour wavelengths: that provides more information than an RGB camera. Earlier in the study, a spectral camera with hundreds of wavelength bands was also used. The amount of data provided by this camera was too large for real-time disease detection, but it could be used to select meaningful bands for the multi-spectral camera.

The research into the cameras will be completed by mid-2022, when WUR will publish a 'proof of concept'. OPTIMA is funded by the European Horizon 2020 programme. More information can be found on the dedicated website.