Insect Pest Detection and Identification Using YOLOv8 on Tomato Crops

Abstract

Agriculture is pivotal to Cyprus’s economy, contributing significantly to the GDP and providing essential benefits in rural development, food security and environmental sustainability. However, insect pests pose a significant threat to productivity. This research investigates the efficacy of the YOLOv8 model in detecting and classifying insect pests on tomato plants, aiming to enhance the field of precision agriculture. The experiments evaluated YOLOv8n,YOLOv8s and YOLOv8m, addressing their performance based on precision, recall, mAP50 and mAP50-95 metrics. Results indicated that that the models performed well in detecting larger insects, and struggled when faced with smaller, less distinct insects. This study highlights the need to address data availability, which hinders work done in this field of research.