Resnet-based modified red deer optimization with DLCNN classifier for plant disease identification and classification

Abstract

The manual inspections of plant diseases resulted in low accuracy with high time consumption and unable to predict the multiple diseases of plants. To address these difficulties, it is necessary to develop automated systems that are capable of effectively classifying. Therefore, this article presents a customized PDICNet model for plant leaf disease identification and classification. Initially, ResNet-50 is used to extract multiple features from plant leaf images with colour and texture properties. In addition, the modified Red Deer optimization algorithm (MRDOA) is implemented as an optimal feature selection algorithm to obtain optimized and salient features with a reduced size of the MRDOA. Further, a deep learning convolutional neural network (DLCNN) classifier model is utilized to achieve enhanced classification performance. Obtained simulation outcome discloses the superiority of proposed PDICNet model with an accuracy and F1-score of 99.73%, and 99.78%, respectively for PlantVillage dataset and 99.68%, and 99.71% for Rice Plant dataset.