Site-Specific Herbicide Application Using Drones and VRT

This project combines drone imagery, soil data, and weather information to create precise herbicide dose maps and Variable Rate Application files. Using Convolutional Neural Networks on ~500 annotated images, it quantifies weed pressure and crop biomass, generating field-wide density maps. These are integrated with soil and weather data to enable targeted herbicide application, reducing usage by approximately 20% without yield loss, as validated in demo farm trials. The system supports iOS and Android apps for image capture at 6 images per hectare with altitude stabilization via ultrasonic sensors. Developed by Agrovast and SOLVI, it evolved into Plant AI—a standardized, drone-compatible solution for spot-spraying and weed detection in field trials.