Mobile-based Assistive Tool to Identify & Learn Medicinal Herbs

Abstract

Sri Lanka is recognized and valued globally due to its rich heritage of tropical plants, herbs and trees. A need for the valuation of valuable herbs are identified among both Sri Lankans as well as tourists. This paper brings forth a solution in distinguishing medicinal herbs through leaves and flowers using deep learning and image processing algorithms via a mobile application. The proposed mobile application identifies a flower and leaf by its morphological features, such as shape, color, texture. The perspective is to achieve highest accuracy for plant identification using image processing. The proposed model revealed an accuracy of 92.5% in the classification of leaves and flowers. Accuracy of 6 different plants are identified using this method. This application also provides Sinhala virtual assistant which enables user to search herbs using the name, which is popular among people, to obtain information about herbs. The main outcome of the virtual assistant of the research is to develop an information retrieval method on medicinal herbs in a more accurate, easy and efficient way. In addition. this application also provides 3D structure of the selected medicinal herb in augmented reality (AR).