Comparative Analysis of YOLOv8 and YOLOv9 Models for Real-Time Plant Disease Detection in Hydroponics
Abhishek Tripathi, Vinaya Gohokar, Rupali Kute
Abstract
Plant diseases are a significant threat to modern agricultural productivity. Hydroponic systems are also affected for various reasons. Reliable and efficient detection methods are essential for early intervention and management of diseases in hydroponics. This study investigates the use of You Only Look Once (YOLO) models, namely YOLOv8 and YOLOv9, for the detection of plant diseases in a hydroponic environment. A diverse dataset was prepared, comprising images from a hydroponics system setup and the New Plant Disease Image Dataset from Kaggle. Custom annotated images were used to train and test the models and compare their accuracy, processing speed, and robustness in hydroponic systems. The results showed that YOLOv9 is slightly better than YOLOv8 in terms of detection accuracy, as it achieved 88.38% compared to 87.22%, respectively. YOLOv8 requires less computational resources and takes relatively less time than YOLOv9 for real-time plant disease detection. Therefore, it is recommended for portable devices.