PJB-2024-164
An Enhanced Method for Agricultural Crop Disease Identification by applying Ontology through Deep Leaning Classification
Rajeswari V
Abstract
The agricultural industry contributes most to expanding economies and populations and is essential to the production of high-quality food. Plant diseases have the potential to completely eradicate species diversity and produce large losses in food production. Early detection of plant diseases through it is possible to lower costs and raise the quality of food production by utilising automated or precise detection methods. Hence, this paper utilized the PlantVillage dataset for efficient plant disease identification. Firstly, data preprocessing is done using resizing and normalization. The ShuffleNet network is used for feature extraction process. It has been shown that the accuracy of classifying plant images is significantly increased by integrating ontologies and semantic relationships. To increase the accuracy of plant image classification, this work has created an ensemble method that combines a hybrid ontological bagging algorithm with convolutional neural network (CNN) models. The ontological bagging idea minimises the classifiers' error propagation by learning weak discriminative characteristics across several learning instances. This research used GhostNet module for classification process. This paper presents the design of Tasmanian Devil Optimisation (TDO), a novel hyperparameter tuning bio-inspired metaheuristic algorithm that emulates the natural behaviour of Tasmanian devils. The Tasmanian devil has two ways to eat: it can attack live prey or eat dead animals' carrion is the primary source of inspiration for TDO. Experiments demonstrated that the recommended model outperformed the most advanced models with a higher classification accuracy of 99.91%.
To Cite this article:
Download
