Enhancing Rice Plant Disease Recognition and Classification Using Modified Sand Cat Swarm Optimization with Deep Learning

Abstract

Rice plant diseases play a critical challenge to agricultural productivity and food safety. Timely and accurate recognition and classification of these ailments are vital for efficient management of the disease. Classifying and recognizing rice plant disease by implementing Deep Learning (DL) has emerged as a powerful approach to tackle the challenges associated with automated disease diagnosis in rice crops. DL, a subfield of artificial intelligence, concentrates to train neural networks with several layers for automated learning of the complex patterns and illustrations from data. In the context of rice plant diseases, DL methods can effectually extract meaningful features from images and accurately classify them into different disease categories.  Therefore, this study introduces a new Modified Sand Cat Swarm Optimization with Deep Learning based Rice Plant Disease Detection and Classification (MSCSO-DLRPDC) technique. The main objective of the MSCSO-DLRPDC technique focalize on the automated classification and recognition of rice plant ailments. To achieve this, the MSCSO-DLRPDC methodology involves two levels of pre-processing such as median filter-based noise removal and CLAHE-based contrast enhancement. Besides, Multi-Layer ShuffleNet with Depthwise Separable Convolution (MLS-DSC) methodology is utilized for feature extraction purposes. Moreover, the Multi-Head Attention-based Long Short-Term Memory (MHA-LSTM) methodology is utilized for the process of rice plant disease detection. At last, the MSCSO method is utilized for the tuning process of the MHA-LSTM approach. The MSCSO approach inspired by the collective behaviour of sand cats and the mutation operator, is implemented for optimizing the parameters of the MHA-LSTM network. To demonstrate the enhanced accomplishment of the MSCSO-DLRPDC method, a broad set of simulations were carried out. The extensive outputs show the greater accomplishment of the MSCSO-DLRPDC method over other methods. The proposed approach has the capability in assisting farmers and agricultural stakeholders in effectively managing rice plant diseases, contributing to improved crop yield and sustainable agricultural practices.