Crafting a Specific Deep Network for Real-Time Identification of Ayurvedic Plants

Plants play vital role for existence of living being specially humans as they rely for food, medicine and for many other needs. Plant-based medicine is an age-old science practiced in many countries. Use of plant-based medicine is considered safer compared to chemical-based medicine for humans because it comprises of natural ingredients. Planet earth is blessed with plenty plant species having medicinal values. However, current generation have lack of knowledge of these medicinal plants. Hence there is a requirement for automated identification of medicinal plants to use them as medicine. In the present work, automated classifying system for identification of medicinal leaves is designed using deep learning approach. Further for real-time usage of the developed classifying system, Android based cell phone application is developed. Medicinal values of the identified leaves are also displayed on the cell phone screen. The dataset required for training the deep network is acquired in the Southern part of Karnataka, India. The system identifies eight types of medicinal leaves with an average accuracy of 99%. Such an automated system will help people associated with ayurvedic medicine, botanists and also common people for using herbs as medicine.