Tomato Disease Detection from Tomato Leaf Images Using CNN-Based Feature Extraction, Feature Selection with Whale Optimization Algorithm, and SVM Classifier

Tomatoes, a popular crop, thrive on nearly any soil with good drainage. Identifying tomato diseases early on is crucial to maintaining quality and yield. This article proposes a hybrid model approach for early tomato disease detection from tomato leaf images. With the applied hybrid model approach, feature extraction was accomplished by employing a lightweight CNN model on images. These extracted features were then optimized by Whale optimization algorithm and utilized to train the SVM classifier for image classification. Validation of the test set utilizing Accuracy, Recall, Precision, and F1 score showed the proposed method achieved an accuracy score of 98.46%. Furthermore, the accuracy improvement obtained was 3.53% when using the proposed compared to the base CNN model. The findings validated that the proposed method outperforms state-of-the-art tomato disease detection methods with high accuracy and a lightweight model.