Deep Learning-Based Detection of Fungal Diseases in Apple Plants Using YOLOv8 Algorithm

Over the last decade, the identification of plant illnesses by automatic means has become a challenging field. In recent times, computer vision has made it feasible to identify diseases quickly and accurately, which has enhanced yield by allowing for the necessary treatment. Graph Neural Networks (GNNs), Convolutional Neural Networks (CNNs), Faster-CNN, Mask R-CNN, Siamese Networks, are the models built on deep learning have been applied to diagnose and cure crop disease. These models accomplish well when it comes to identifying plant diseases, but they have trouble with complex field images and frames with several leaves. The goal of this study is to enhance their application in real-world agriculture, where problems with intricate images and multiple leaves are frequent. This paper uses a YOLOv8 model to identify fungal diseases that affect apple leaves, including cedar rust, scab leaves, and black rot. The dataset used for the implementation was gathered from a Kaggle of 540 images, and the model obtained 86% mAP, outperforming the existing YOLOX model which has achieved 73^%mAP.