LTECP: Experimental Analysis of Medicinal Plant Recognition System Based on Learning Assisted Texture Evaluation and Classification Principle

The experimental analysis of medicinal plant recognition systems based on the Learning-assisted Texture Evaluation and Classification Principle (LTECP) aimed to develop an efficient system for identifying medicinal plants using machine learning and deep learning models. The dataset consisted of multispectral and digital images of various medicinal plant leaves. Image preprocessing, feature extraction using texture and geometric features, and feature optimization through chisquare selection were integral parts of the system. Multiple classification models, including Random Forest, Support Vector Machines (SVM), and Convolutional Neural Networks (CNN), were tested. The CNN models, specifically InceptionV3, ResNet50, and VGG-16, significantly outperformed traditional machine learning models. The InceptionV3 architecture achieved the highest accuracy of 98.20%, while ResNet50 and VGG-16 reached accuracies of 97.00% and 96.50%, respectively. The best-performing models demonstrated high precision, recall, and F1-scores across different plant species, with minimal misclassification errors. Feature optimization improved model performance by focusing on the most relevant data points, increasing accuracy across all models by up to 5%. The results highlight the potential of CNN architectures for accurate medicinal plant recognition.