Identification of Plant Leaves having Anti-Diabetic Property using Machine Learning

Abstract

About 1200 medicinal plants have been used in Ayurveda, Unani and Chinese medicines for the management of diabetes. Diabetes is an endocrine disorder where glucose levels rise in blood due to lack of insulin production by pancreas. Identification and detection of these plants manually can be extremely tedious and time consuming thus; using machine learning is more beneficial and promising. Due to improved capacity of machine learning to acquire, manage, and store extremely vast volumes of data, machine learning is being trained to be applied to identify the plants, their phenotype using images of plants and their disease. This research study has proposed the application of machine learning for the identification of leaves of plants possessing anti-diabetic property. Here, the machine learning algorithms are applied for the detection of leaves of some anti-diabetic plants namely Basella alba, Moringa oleifera, Fenugreek, Psidium guajava, Hibiscus rosa sinesis. In the proposed experiment, the highest accuracy of about 99.4% was achieved by using a combination of Neural Network and Logistic Regression. The proposed model effectively classifies the plant images with high accuracy.