This smartphone app integrates a deep learning model to detect and map two-spotted spider mite (TSSM) populations in strawberries, enabling growers to precisely and accurately treat them. Every year, thousands of strawberry growers face significant yield and quality losses due to two-spotted spider mite (TSSM) infestations, one of the most destructive pests in berry production globally. The TSSM feed on the leaves and influence plant photosynthesis. To manage the infestations, growers need to manually count the number of TSSM under the leaves under magnifiers or microscopes to determine the amount of pesticide needed. This approach is labor-intensive, inconsistent, and impractical for large-scale operations.
Researchers at the University of Florida have developed a smartphone application that automates the detection, counting, and spatial mapping of TSSM populations in real-time using a deep learning model. The system is trained on 2,713 macro-lens images collected from various smartphones and uses a convolutional neural network to accurately identify and quantify mites in captured images. The app counts TSSMs, records the GPS coordinates, and stores the data. It offers a geo-referenced visualization of sampling points and uses spatial interpolation to generate heatmaps of mite populations across the field.
A field-deployable mobile tool for real-time pest monitoring and spatial distribution mapping in strawberry fields, supporting precise miticide application and biocontrol deployment
This technology enables real-time detection and spatial mapping of two-spotted spider mites (TSSM) in strawberry crops using a smartphone-based deep learning system. The core of the system is a convolutional neural network trained on over 2,700 macro images of TSSM captured with consumer smartphones and 25x macro lenses. The application allows growers to capture close-up images of strawberry leaves, automatically detect and count mite populations, and log GPS coordinates alongside pest data.