Classification of Rice Leaves Diseases by Deep CNN-Transfer Learning Approach for Improved Rice Agriculture

Abstract

Rice leaves may suffer serious impacts such as low production or yield of the respective products if necessary precautions are not taken. Therefore, to ensure the healthy and normal growth of the rice plants, early diagnosis of any disease and application of the necessary therapy to the damaged plants are paramount. Because manual disease diagnosis requires a lot of time and effort, an effective automated method is required for early disease diagnosis. As a result, this study presents a deep learning-based solution to the aforementioned issue for the automated detection of three plant diseases: leaf smut, bacterial leaf blight, and brown spot that frequently affect rice plants. The transfer learning with VGGNet convolutional neural network (CNN), which was pre-trained on a sizable Imagenet dataset, was used in this study to effectively classify the illnesses of rice leaves. A number of cutting-edge classifiers, including Support Vector Machine (SVM), k-nearest neighbour (kNN), Convolutional Neural Network (CNN), Random Forest, and Decision Tree, are used to compare the performance of the proposed framework. The results demonstrate that the proposed CNN-transfer learning framework outperforms other classifiers with a mean accuracy of 97.22% in the 5-fold cross-validation.