Innovative Approach to Precision Botany: Hybrid Learning for Superior Medicinal Plant Identification

Modern technology has improved upon ancient approaches, making them more resilient and scalable, recognizing the need for precision and effectiveness in identifying medicinal plants. The integration of traditional botanical knowledge with contemporary technology in the 21st century improves our understanding of the medicinal properties of plants, which have long been vital to civilizations and life on Earth. This study describes an innovative plant classification technique that utilizes classic methods with sophisticated support vector machines (SVMs) for classification and convolutional neural networks (CNNs) for feature extraction. The goal is to transform the identification of medicinal plants, paving the path for novel medical and pharmaceutical innovations. Accurate identification of therapeutic plants is critical to medicinal science, medicine, and botany. To ensure resilience, we trained, tested, and validated our hybrid CNN-SVM model on two distinct datasets: "BDMediLeaves" and "NSSmediplants". Where the proposed dataset "NSSmediplants" comprises both real and improved plant photographs, all expertly photographed in natural light against appropriate backdrops. The recognition rates obtained were 94.11% and 99.5%, respectively. Also used k-fold validation technique to ensure the effectiveness of our noble hybrid model. By integrating the validated novel model we developed a smartphone application. This work advances the efficiency and accuracy of image-based plant categorization using CNNs and SVMs, facilitating further research and practical applications in this critical field.