Recognition of ayurvedic medicinal plants from leaves: A computer vision approach

摘要

Plants are an indispensable part of our ecosystem and India has a long history of using plants as a source of medicines. Since the advent of modern allopathic medicine, the use of traditional medicine declined to a considerable extent. However, in recent years, traditional medicine has made a comeback for a variety of reasons like they are inexpensive, nontoxic and does not impact any side effect. Different kind of medicinal plant species are available on earth but it is very difficult to identify the plant. Considerable knowledge accumulated by the villagers and tribal on medicine from plants remains unknown to the scientists and urban people. This kind of knowledge is usually handed down through generations. Our immediate concern is to preserve this knowledge in digital form through the concepts of machine learning, pattern recognition and computer vision. A machine can identify a medicinal plant through the features extracted from the leaf images, together with a classification algorithm. This paper proposes a computer vision approach for the recognition of ayurvedic medicinal plant species found in Western Ghats of India. The proposed system uses a combination of SURF and HOG features extracted from leaf images and a classification using k-NN classifier. Our experiments show results which seem to be sufficient for building apps for real life use.