A Convolutional Neural Network Approach for Automated Medicinal Flora Recognition

Abstract

Medicinal plants are integral to both traditional healing practices and modern pharmacological research. However, accurate identification of these plants remains a persistent challenge, particularly for individuals without botanical expertise. This paper presents a deep learning-based system for automatic medicinal plant identification and information retrieval. The proposed solution employs a Convolutional Neural Network (CNN) architecture, specifically MobileNetV2, to classify plant species from user-submitted images with a classification accuracy of 75% with certain species reaching up to 93.5% accuracy. Upon successful identification, the system provides comprehensive medicinal information, including the plant’s scientific name, therapeutic uses, and health benefits. A user-friendly web application is developed using Streamlit, enabling real-time interaction, while MongoDB is used for storing user interaction and metadata. This system is intended to assist students, researchers, farmers, and healthcare professionals by bridging the gap between plant identification and medicinal knowledge through an accessible, automated platform.