Early Diagnosis of Tomato Diseases Using Machine Learning to Reduce Pesticide Application

Abstract

Pesticides, which are widely used today to increase yield and prevent diseases in agricultural products, have negative effects on human health and the environment. Due to these effects, efforts to restrict pesticide use gained momentum with environmental protection movements starting in the 1970s. Strict controls and licensing processes have been introduced for pesticide production to prevent their unconscious use. The aim of this study is to classify diseases observed in tomato plants, thereby preparing a scientific groundwork for reducing pesticide use. For this purpose, a dataset comprising a total of 13,883 records of tomato plants belonging to six different classes was utilized. For the classification of the data, the following Machine Learning methods were employed: Artificial Neural Network (ANN), K-Nearest Neighbor (KNN), Support Vector Machine (SVM), and Logistic Regression (LR) algorithms. Model performance was evaluated using metrics such as accuracy, precision, recall, and F1 Score, along with the confusion matrix and ROC analysis. The classification success rates obtained from model training and testing were determined as: ANN 100.0%, KNN 99.8%, SVM 99.9%, and LR 99.9%. The results indicate that all models achieved high and very similar success rates. In conclusion, it is expected that all classification models used can be successfully utilized for the reliable diagnosis of tomato plant diseases, and this capability is projected to make significant contributions to organic farming applications.