The classification of plant species used in herbal medicine is one of the most popular applications of machine learning, deep learning and image processing. There are countless species of plants, some of which have therapeutic benefits, in the world. Herbal, Ayurvedic, and Traditional Medicine all make heavy use of medicinal herbs in their production. Among the different methods for identifying plants and herbs is leaf identification, thus it is essential to accurately examine and classify plant leaf. The leaves are sorted into groups based on their shape and size. The automatic identification system for medicinal plants described in this paper will help people in the community learn more about these plants. This paper presents a CNN model to classify 45 different classes of medicinal leaf, and found accuracy of 93.53% after 50 number of epochs.