Medicinal Plants Identification and Their Medicinal Properties Detection with Chatbot Integration

Traditional medicinal plant knowledge is fading, and existing digital identification tools lack scalability and detailed information. This paper presents a comprehensive platform for medicinal plant identification and retrieval. We use an Xception Convolutional Neural Network (CNN) to classify 201 distinct categories (200 medicinal plants plus a “Not-a-Medicinal-plant” class) with high scalability. The model is trained on preprocessed images (resized and normalized) and achieves 98.5% validation accuracy and an F1 Score of 0.96.Beyond classification, the system’s key innovation is its integration of a multilingual (English, Hindi, Marathi), voice-enabled chatbot powered by the Gemini API. This chatbot dynamically provides detailed medicinal properties and uses without a pre-built database, directly addressing the information gap in other systems. This integrated solution bridges computer vision and NLP, offering a scalable and accessible tool for researchers, students, and the public.