Improved Identification of Medicinal Plants Utilizing ParameterOptimized SVM inside a PSO-Driven Cascaded Framework

Abstract

The primary objective of this work's extension is to improve medicinal plant classification by parameter modification of the Support Vector Machine (SVM) in a PSO-optimized cascaded network. Deep features obtained with ResNet50 are enhanced using Particle Swarm Optimization (PSO) to choose the most relevant attributes. The customized SVM model significantly improves classification accuracy and reduces misclassification when compared to alternative methods. Experimental results show that the enhanced SVM outperforms the others, with an accuracy of 99.75%. This approach ensures a more reliable, efficient, and scalable automated medicinal plant identification system.