Detection of Diseases on Apple Leaves with Deep Learning

In agricultural production, timely and accurate detection of diseases is of critical importance in terms of preventing product losses, supporting environmentally friendly agricultural practices and increasing productivity. In this study, it is aimed to increase model performance by performing finetuning with deep learning based models for early detection and accurate classification of diseases on apple leaves. Due to the time-consuming and expert nature of traditional diagnosis methods, this study proposes an automatic diagnosis system that can provide fast, practical and reliable results to farmers. Thus, production efficiency will be increased and agricultural sustainability will be supported. Four different DL models, namely Vision Transformer (ViT), EfficientNetB0, InceptionV3 and ResNet50, were used in the study. As a result of the research, the model with the highest performance was ViT with an accuracy rate of 97%. The high accuracy rate of this model shows a significant success in correctly identifying leaf diseases.