Early detection of plant diseases

Distr@l Projekt: Frühzeitige Erkennung von Pflanzenkrankheiten
In the agricultural sector, a so-called large-scale crop classification offers for the first time the possibility to check each individual field for vitality, yield, damage or processing steps without having to have contacted the individual farmer or been on site. In this context, the early detection of plant diseases is being developed in this Distr@l project and made accessible to users. In the project, an automated evaluation of satellite data will be developed that enables early detection of plant diseases already during an early growth stage of the plants. Ideally, potential diseases in the most important crops such as winter wheat and winter oilseed rape will already be detected during winter dormancy. The models are fed with satellite data, in particular from the European Copernicus mission, and build on the previous research results and products of CORAmaps, namely the large-scale determination of arable plants, the detection of their vitality and the precise recognition of field boundaries. The weekly availability of data from various Earth observation satellites allows the approach to be easily scaled up for the entire surface of the Earth, even beyond the borders of Europe, and is therefore seen as a basic product for global mapping of plant diseases.
The knowledge gained from this is particularly interesting for farmers for the economical and environmentally friendly use of plant protection products. Accordingly, the diseases are not detected by local inspections and visual observation, as was previously the case, but are recorded by means of areal and weekly measurements by satellite sensors of the latest generation, which cover numerous ranges of the electromagnetic spectrum. In addition, the prediction of potential occurrences of pest types is made possible by measuring various environmental parameters through satellite data and learning a probability of occurrence based on deep neural network technology.

This project is funded by

Hessische Staatskanzlei logo
distral logo