Automatic Infestation Alert for Leafhopper using Drone Imagery
- To develop a Leafhopper Symptom Early Detection Pipeline, using drone-acquired multispectral data to automatically detect and classify the symptom severity class of each vine;
- To develop a drone flight recommendation model based on InPP risk prediction model to indicate whether drones should be deployed in two weeks notice;
- To develop an mobile phone alert communication system that informs the farmers of: risk prediction of leafhopper symptom severity; drone flight timing recommendation; early symptoms of leafhopper infestation in vineyards.
Challenges and how they will be addressed
Technical challenges:
- Would a drone mounted multispectral camera be able to detect critical leafhopper severity symptoms?
- Would the current InPP risk prediction model define the time for flying the drone and predict pest symptoms accurately?
Data collection and model training will be critical as well as the integration between the air and ground level monitoring.
Societal challenge:
- Would a small-medium company invest and embrace the proposed digital technology?
InPP’s approach is to be inclusive from the start, describing the technology and letting the farmer familiarise with it.
Business challenge:
- Would the project deliver a tool in TRL5 that convinces the wine producers and drone companies?
Tech components and data
The project will focus on the dominant red varieties in Alentejo region: Alicante Bouschet, Trincadeira and Aragonês. In order to allow us to connect ground and air-level monitoring, the main following datasets will be collected, from 6 different plots (weekly, June-September):
- aerial RGB and multispectral images (validated with field-collected hyperspectral data);
- Chromotropic trap images to obtain counts of adult leafhoppers;
- farming practices, specifically insecticide applications;
- Visual identification of symptom severity classes at the leaf level;
- meteorological data.
Multiple tech components will be used to generate, analyse and develop predictive models from the above data. Namely, the following components will be used to collect data: a DJI Mavic 3M drone (equipped with RGB and multispectral cameras); a field spectroradiometer; weather stations; mobile applications. Image analysis algorithms, machine and deep learning methods and a relational database will form the core analytical components of the system.
Expected outcomes
1
Well-annotated datasets of air and ground-level monitoring of leafhopper-caused symptoms and population dynamics
2
Pipeline for the automated pre-processing of drone-acquired multispectral data
3
AI-based Leafhopper Symptom Early Detection model (per vine), based on drone-acquired data
4
Flight recommendation alert model, based on a risk prediction model of symptom severity
5
Automated communication system to inform the farmer on the outputs of the different models that are part of AI4Leafhopper
About InnovPlantProtect
| Full name | InnovPlantProtect – Associação |
| Country | Portugal |
| Website | https://iplantprotect.pt/ |
| Contact Persons | iLaria Marengo (drone monitoring), Ricardo Ramiro (data analysis) |
| Contact’s email | , |
InnovPlantProtect (InPP) is a Collaborative Laboratory (CoLab) which started its activities in January 2020. It was created with the aim to develop innovative bio-based solutions to protect Mediterranean plants from pests and diseases and to provide diagnostic and monitoring services, thus contributing to the sustainability of agro-forestry systems. A primary focus of InPP is plant protection against emerging pests and diseases associated with climate change, the harmful effects of which are already visible in Portugal and in many other countries. InPP is a multi-cultural organisation of 44 employees with MSc and PhD qualifications. InPP works with the public and private sectors as well as with other CoLabs and Universities thanks to a broad range of services and funded projects (at National and European levels), which help strengthen the InPP network.
Stay Connected
We appreciate your feedback.
Join our newsletter
Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or Research Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.