A Technological Perspective on Traditional Medicine: Classification of Plant Species with Machine Learning

Abstract

The use of medicinal plants with rich chemical content has been one of the methods used for centuries to solve the basic health problems for both developed and underdeveloped societies. It is known that this method is used effectively in Far East countries. The level of success that modern medicine has reached in protecting and treating human health has led to the oblivion of traditional medicine and therefore of what is known about medicinal plants. It seems that traditional medicine will continue to exist in our lives for a while due to some reasons such as the rapid increase in the world population, the inability of every society to reach the drugs at the desired level and the sometimes negative effects of drugs. For this reason, diagnosing and correctly identifying plants with medicinal contents is a very vital situation for people who will use these plants for therapeutic purposes. Knowing the ingredients of any herb is the work of a specialist. An ordinary person's misidentification and use of a plant for therapeutic purposes can cause negative consequences. For this reason, the aim of this study is to identify any plant with the help of its visual morphological features such as the leaf shape, color or odor and to make an accurate description of its medicinal content by image processing and machine learning method. For this, medicinal plants could be identified with 90% accuracy by using the KAGGLE database, which contains 100 images of leaves and flowers for each of the two plants.