Comparative analysis of paddy leaf diseases sensing with a hybrid convolutional neural network model

Abstract

The major food in south India is Paddy. Early detection of leaf disease can increase the production of crops. In the agricultural field, farmers can control the spreading of leaf disease with the help of the modeling of automated disease classification. In comparison to the manual approach to paddy leaf infection detection via visual examination, automated identification of rice illnesses using an image of the paddy leaf may be advantageous. Deep learning, a highly well-liked and effective machine learning method, has recently demonstrated considerable promise in the categorization task of images. In this work, Contrast Limited Adaptive Histogram Equalization (CLAHE) pre-processes the augmented images of paddy leaves. The Gray level co-occurrence metrics (GLCM) model then extracts the features from an image of the pre-processed paddy leaves. The five disease classes of paddy leaves were finally identified using the hybrid CNN model as follows: Bacterial Leaf Blight (BLB), Sheath Rot (SR), Narrow Brown Leaf Spot (NBLS), Brown Spot (BS) and Leaf Smut (LS). MATLAB platforms handle implementation works. While compared to the previous approaches, the proposed model reveals the superiority of paddy leaves classification.