CP6 & CP7: Lab-to-field upscaling of biologicals and AI-assisted pathogen monitoring

In a nutshell

What?

We are developing a monitoring system for the early detection of plant diseases and other stress factors in the field with the help of AI-supported pathogen monitoring.

We are also testing the effect of various biological antagonists against two major plant diseases, Fusarium Head Blight on wheat and Sclerotinia Stem Rot on soybean, as an alternative to chemical plant protection.

Why?

Early detection of plant diseases and stress factors is important in order to be able to initiate effective countermeasures as quickly and effectively as possible. As chemical plant protection products have harmful effects on the environment and can remain as residues in food, the establishment of alternative biological control measures is sensible and necessary.

How?

The starting point is the hyperspectral data already obtained during funding period 1. With the help of this data, an AI process based on deep learning was developed, which serves as the technological basis for the first automated detection of pathogens in the plant population. Further data is continuously being systematically collected in order to further train the deep learning-based algorithms and obtain increasingly robust pathogen recognition under different environmental conditions.

In the case of biological antagonists, the second funding phase will increasingly focus on fungal species. The aim is to achieve greater flexibility and effectiveness in the field using different application methods (location, time) and formulations.

 

 

 

CP6:
Dep. Phytopathology (360a)

Otto-Sander-Str. 5
70599 Stuttgart

CP7:
Dep. of Artificial Intelligence in Agriculture (440g)

Garbenstr. 9
70599 Stuttgart

in planning stage


Subproject Team

Prof. Dr. Ralf Vögele
Subproject Leader

Prof. Dr. Ralf Vögele

Jun.-Prof. Dr. Anthony Stein
Subproject Leader

Jun.-Prof. Dr. Anthony Stein


In order to realize an ecological and economical agriculture, efficient use of resources in plant protection is necessary. The detection of plant diseases in the field is complex, as fungi grow on or in plant tissue and are not visible to the bare eye at the beginning of an infection (Hallmann et al. 2019). Early detection is essential in order to minimize yield losses through rapid and appropriate countermeasures.

This also includes the use of biological antagonists, with the potential to replace chemical plant protection products in the medium to long term without the problem of chemical residues or the development of drug resistances. To achieve this, it is necessary to quantify the pathogens, also using AI-based remote sensing, in the plant population at an early stage and to initiate control measures promptly. However, further research into the development of new biological antagonists that improve efficacy in the field through innovative formulations is required for practical use in the field. The right time, type and location of application also have a major influence on the effectiveness of biological control (Bejarano & Puopolo, 2020).

The aim of the joint research project for the second funding period is to utilize the successful application of specifically formulated biological control agents with the aid of AI-supported pathogen monitoring for the further development of an innovative plant protection system in a NocsPS cultivation system. This should enable the early detection of stress factors and the timely initiation of appropriate countermeasures.

In order to achieve greater flexibility and effectiveness of the biological antagonists, it is planned to compare different application methods (foliar treatment, soil application, seed dressing) and to test various formulations.

The work program of the Phytopathology and Artificial Intelligence in Agricultural Engineering departments is divided into two overarching work packages:

WP1: Establishment of preparations with biological antagonists for the control of ear fusarium and white stem borer

  • Clarification of further indirect and direct modes of action of biological antagonists, as well as their performance under field conditions (rain, UV radiation)
  • Optimization of application methods and locations using AI-based analyses

WP2: Development of methods for AI-based plant pathogen detection in soy and wheat

  • Development of an automated data acquisition under controlled conditions in the greenhouse of multi-modal spectral images for the training of the targeted robust deep learning models
  • Working out the continuous need for optimization in order to continually incorporate suitable adaptations from the rapidly developing state of AI research.
  • Investigation of the transferability of the developed AI methods from the lab-scale, over simulated conditions in the green house, up to real field conditions in order to ensure later feasibility in practice.