Optimize
flight planning and acquisition of hyperspectral images with drones.
Improve
the quality of UAV-based hyperspectral images and automate the workflow.
Enhance
performance of in-house ML algorithms to early identify the plants affected by the disease with rigorous validation with field measurements.
Commercialize
the monitoring services and the decision support system.
HYGRI
HYGRI aimed to acquire and use UAV-based hyperspectral imagery to monitor and detect plant pathogens in fruit trees and to visualize the information in a user-friendly decision support system. Existing know-how at AVT-ASI provides the basis for operating the UAV system and detecting plants affected by a number of diseases, for example flavescence dorée in vineyards, and for analyzing plant health status, but advances were needed to solve challenges in the acquisition and processing phases and in the choice of the machine learning approach.
DJI Matrice 600 Pro, DJI Mavic 2 Enterprise Dual
AFX10 hyperspectral sensor
Gimbal
GPS STONEX S8 Plus GNSS
Drone Data analytics models
Model 1: Spectral analysis
Model 2: Classification of plants based on health status
Model 3: Multi-temporal analysis
Model 4: Thermal images analysis"
Outcomes
Correspondence between vegetation indices and ground sensors (TER)
Classification of stressed / unstressed plants
Advances with MAPPIS visualization tool
Challenges Faced
UAV-flight execution
Local people were suspicious (use of drones during war was influencing)
Ιnformative material used, contact details in the company and farms
Staff was wearing uniforms to be recognized
Topography of terrain, presence of obstacles
Flight plans modified
Image classification with respect to health status
Number of samples of unhealthy plants not sufficient to train the model
Impact (Socio-economic & Environmental)
Socio-economic impact
Lower exposure to chemicals for farmers
Support for sustainable land management
Enhancement of efficiency in agriculture and costs reduction
Positive effects on rural innovation
Ιncreased trust in digital technologies
Environmental impact
Water resources optimization
Pesticide reduction
Early detection of diseases
Wide areas monitoring at low impact
Improved crop management
Optimization of resource use