Identification of Different Medicinal Plants through Image Processing Using CNN-GMM

Medicinal plants are commonly used as therapeutic agents in traditional medicine and demand for natural health products is increasing worldwide. However, identification of medicinal plants by inspection is a source of error, which can result in the mistaken view, and the treatment has serious consequences. We have proposed a new hybrid deep learning model called CNN-GMM for the automatic prediction and classification of medicinal leaves. The system also reaches increased accuracy and robustness by using Convolutional Neural Networks for feature extraction, and Gaussian Mixture Models for probabilistic clustering. The model achieved high-performance metrics for accuracy, precision, recall, and F1-scores and was trained using a Segmented Medicinal leaf dataset, which consisted of 1835 images of several plant species present in India. Output also includes supporting metadata (e.g., plant names, uses, and distribution) to help elucidate their medicinal use. And while building upon and improving the process of sourcing, identifying, testing, and characterizing plants provides a hybrid model that has promising implications in terms of botany, pharmacology, and the study of medicinal plants, potentially allowing for better conservation and use of these resources.