Prospect FD : development of an aid decision-making tool for the prospection of vine Flavescence Dorée – Prospect FD

To achieve the objectives set, the project is structured in 4 tasks :

Task n ° 1: Realization and adaptation on a harvesting machine of an image acquisition and storage system.

This lot is based on the use of high resolution cameras in the visible domain. The system developed should allow observation of the vine in its training plan, from the inter row.

Task n ° 2: Development of image processing algorithms and construction of risk indicators

The approaches considered for the detection of these symptoms fall into two broad categories. The first category consists of identifying radiometric, geometric or textural descriptors capable of discriminating between the different objects present in the image and using them as inputs to segmentation or automatic classification algorithms.
The second category of approaches, in the field of artificial intelligence, show greater generalization capacities. They are based on so-called deep learning techniques and aim, like the previous ones, at the classification, segmentation or detection of objects of interest in images.

Task n ° 3: Creation of a Decision Support Tool to optimize prospecting campaigns

The first work will consist in developing an infrastructure making it possible to easily manage and use the images acquired in the vineyard.
A second job will consist of collecting, managing and prioritizing information to optimize the fight against FD.
Finally, Decision Rules combining all of this information will be created in order to process the available information and to propose an optimized prospecting strategy based on all of this information.


Task n ° 4: Field testing and validation of the tools developed

Indispensable, this task aims both to acquire the basic data for the construction of tools and to evaluate them in the vineyard.

The year 2020 made it possible to acquire images in the vineyard of white and red grape varieties. The analysis and processing of the images will be able to begin ...

Ongoing project...

Ongoing project...

The mandatory control measure against the Flavescence Dorée is opertional since many years in the French vineyard and need the uses of inscecticides in huge quantity each year. When the prospection is important we can see a diminutuion of the contamination, but when only 10% of the vineyard is prospected each year (more is impossible due to a low added value on the vine) the contamination is increasing.

Prospect FD project aim to combine différents methods (imagery, artificial intelligence, information system, precision viticulture and human expertise) to detect the Flavescence Dorée. It relies (i) on a choise of embedded sensors for the harvesting machine, (ii) on data infrastructure to centralise, to compute and to make available the acquired informations, (iii) on a geographic tool (aid making-decision tool) presenting the informations (i.e. pictures, disease risk mapping) to human expertise. The objective is to capture datas while the harvesting machine is passign in each vine rows, close to the vegetation and during an optimal period to watch the symptoms and to spot the potential sick plants. The provided maps can be used to check Flavescence Dorée presence during the year N, or can be an information for the aid making-decision tool to help the prospection team during the year N+1 to check the plots where the conatmination risk is heavy.