A machine learning approach for the identification of olive fruit fly in Greece

Abstract

Contemporary agriculture faces critical challenges to maintain a future that meets global food demand. Precise and early detection of plantations’ pest and disease threats is crucial for controlling their spread, maintaining production quality and volume, minimizing costs, and reducing trade disruptions, sometimes even lessening human health risks. Pest management in agriculture benefits significantly from the application of deep learning (DL) techniques for more efficient detection and monitoring, overcoming the inefficiencies of traditional labor-intensive methods. This study develops a convolutional neural network (CNN) and benchmarks it against state-of-the-art (SOTA) DL models to identify the primary threat to olive trees, Bactrocera oleae (also known as Dacus). Using a data set composed of images that span 102 insect categories, CNN demonstrated a high accuracy of 96. 32% to distinguish Dacus from other insect species.