Leaf recognition and classification algorithm to be used by indigenous medicine

In Sri Lanka there is a lot of emphasis on herbal and indigenous medicine. Majority prefer Ayurvedic medicine to Western medicine due to various factors such as less or rather no side effects. The fact that Ayurvedic medical system stood the test of time sums up to its worth as an alternative course of treatment for various ailments. However, a major drawback in the practice of indigenous medicine is the difficulty in finding ingredients. We developed a system that recognises the leaves of the herbs used for Ayurvedic treatment. The system takes the leaves of the trees as inputs and outputs the name of the tree or plant. We used a combination of image processing techniques and neural networking to train a model and perform pattern matching to recognise the leaf with a pre-set level of confidence for its accuracy. As a supplementary feature, the system is also capable of providing a description of the identified herb´s medicinal usage in reference to illnesses by going through feed data.