Digital Holography and artificial intelligence for real-time detection and identification of pathogenic airborne spores

Abstract

Ever-growing concerns and governmental restrictions related to the use of pesticides in modern agriculture has driven the need for more adept decision-making tools to minimize unnecessary treatments whilst still efficiently preventing a spread of infection. To this effect, a network of cost-effective, laser-based holographic detectors were developed and placed in vineyards in Switzerland and France with the objective of detecting and identifying airborne spores of downy and powdery mildew before they have the potential to infect crops. The data collected are remotely sent to a server where image processing techniques and artificial intelligence classify the spores and determine the quantitative intervention thresholds. Knowledge on the quantitative development of fungal diseases has been successfully used to temporally and spatially identify the primary infection of downy mildew which was confirmed by a visual evaluation of symptoms within the parcel. This data coupled with the current risk prediction models provide farmers with a powerful decision-making tool to optimise strategies in the management of grapevine diseases.