Detection of paddy plant diseases using SVM-RFE, AROA with MBi-LSTM model

Abstract

The prime objective of this manuscript is to identify diseases affecting paddy plants, which are an important cause of crop loss. The proposed work have been carried out into four stages: stages first deals with data gathering and pre-processing where data have been gathered from kaggle and in pre-processing quality have been improved. To clean up the green section of the image, a median filter has been applied on an RGB image. Stage second deals with extraction of features where texture and colour features have been extracted from each cleaned image. Stage third deals with selection of features where the most important features have been selected using SVM-RFE, AROA, and both the intersection of SVM-RFE and AROA methods. Last stage fourth deals with the MBi-LSTM classification on the selected features to detect the diseases in the image. The proposed method namely intersection of both SVM-RFE and AROA with MBi-LSTM has shown the ability to detect paddy crop diseases as well.