Diagnosis of tomato leaf disease using OTSU multi-threshold image segmentation-based chimp optimization algorithm and LeNet-5 classifier

Abstract

The ability to diagnose crop diseases is crucial which affects the crop yield and agricultural productivity. The primary area of study for crop disease diagnostics now centres on deep learning techniques. However, deep learning techniques require high computational power, which limits their portability. This paper used the variation of convolution neural network model LeNet-5 for classification and the Otsu multi-thresholding method with an optimization algorithm for the segmentation of the images. The classifier is trained using the Plant Village dataset which contains images of tomato leaves with various types of diseases. This method is highlighted for its high accuracy in disease identification. Additionally, to assess its ability to perform well with new, unseen data, real-time diseased images are tested in the proposed method. This can ensure that the method can effectively generalize beyond the initial dataset it was trained on. The performance using the dataset can be calculated using precision, recall, F1-score, and accuracy. These are compared with three existing approaches Xception, ResNet50, and VGG16 from this comparison the proposed approach gives the best accuracy for classification.