An Intelligent Framework for Rice Plant Disease Recognition and Classification using Deep Learning and Reptile Search Algorithm

Abstract

Rice is one of the majorly utilized primary crops globally, providing sustenance for the worldwide population’s crucial segment. However, if left untreated, rice plants are susceptible to various diseases that can cause substantial yield losses. Hence, precise and timely recognition of rice plant diseases is substantial for effectively managing the disease and ensuring optimal crop production. Therefore, this study concentrates on constructing an Automated Rice Plant Disease Detection utilizing a Reptile Search Algorithm with Deep Learning (ARPDD-RSADL) technique. The purpose of the ARPDD-RSADL technique lies in the accurate and proficient classification of rice plant ailments using a parameter-tuned DL model. To accomplish this, the ARPDD-RSADL technique performs image preprocessing in two stages: Unsharp Masking (UM) based filtering and contrast enhancement. In addition, the ARPDD-RSADL technique designs the Squeeze and Excitation ResNet (SE-ResNet) model for feature extraction purposes. The Stacked Sparse Auto-Encoder (SSAE) model is used with RSA based hyperparameter optimizer for plant disease detection. The design of RSA helps to adjust the parameters relevant to the SSAE model optimally, and it assists in attaining advanced accuracy in detection. An extensive experimentation analysis was accomplished to validate the effectual rice plant disease classification results of the ARPDD-RSADL method. The simulation values portrayed the ARPDD-RSADL method's excellence in diverse evaluation measures. Investigational outcomes illustrate the effectiveness of the proposed model in accurately classifying and recognizing several rice plant diseases.