DEEP LEARNING FRAMEWORK FOR MULTI-CLASS TOMATO LEAF DISEASE DETECTION FROM SEGMENTED IMAGES

Abstract

The diseases that occur in tomatoes impact the production and quality of the crops resulting in losses to the agricultural industry. Disease detection is important for managing crops and maintaining sustainable agricultural operations. The paper proposes a model based on deep learning algorithms which can detect multi-class tomato leaf diseases throughout the usage of segmented images. This paper uses publicly available dataset which includes segmented images of tomato leaves fitting to various classes of disease types such as bacterial spot, early blight, leaf Mold, late blight, septoria leaf spots, spider mites, mosaic virus, target spot, yellow leaf curl virus, and healthy tomato leaves. Image processing methods were performed on the image dataset. The architecture of CNN is used for automatic discrimination of features and classification for disease. Performance of the proposed system was measured based on the standard parameters like accuracy, recall, precision, F1-score, and confusion matrix. Consequences from experiments direct that the deep learning model accurately detects different diseases of tomato leaves with high classification accuracy. Segmentation of the leaf makes the process much easier and permits the model to spotlight on the key regions where the disease can occur. The proposed approach could be considered as an efficient tool for tomato disease detection.