SVETNet: A Deep Feature-Integrated Hybrid Framework for Hierarchical Multi-Output Medicinal Plant Classification

Abstract

The identification of medicinal plants has long been a crucial area in healthcare, agriculture, and traditional medicine systems such as Ayurveda. Traditionally, plant identification has relied on manual observation by experts, which is time-consuming and requires extensive domain knowledge. With the advancement of artificial intelligence and computer vision, automated approaches have emerged to support faster and more accurate plant classification. However, conventional systems often depend on basic image processing techniques and limited feature extraction methods, resulting in lower accuracy and poor generalization when applied to diverse plant datasets. These challenges highlight the need for a more robust and intelligent system capable of analysing complex visual patterns in plant images. This study addresses the problem of accurately identifying medicinal plants from images by proposing an advanced image-based classification framework. The system utilizes InceptionResNetV2 for deep feature extraction, capturing essential visual characteristics such as leaf shape, texture, and structural patterns. These features are then used to train multiple machine learning (ML) models, including Gaussian Naive Bayes (GNB), K-Nearest Neighbors (KNN), Restricted Boltzmann Machine (RBM) with Logistic Regression (LR), and the proposed hybrid Support Vector Extra Trees Network (SVETNet). The SVETNet model combines the strengths of Support Vector Classification (SVC) and ensemble-based Extra Trees (ET) to achieve superior classification performance. Experimental results show that the proposed SVETNet model achieves the highest accuracy of 0.9947 for main class classification and 0.9594 for sub-class classification, outperforming all baseline models.