Evaluation of Support Vector Machine and Binary Convolutional Neural Network for Automatic Medicinal Plant Species Identification

Abstract

Enormous amount of diversified plant species are available in India. Recognition and classification of these species have become a major challenge and an important research field. Though different parts of plants can be used in identifying their genre, leaf is most useful and effective method in classification. Machine learning brings an ideal way to automate this system. A separate dataset is built by collecting 20 different leaf samples available mainly in Southern India. More than 20,000 such samples are collected to build this dataset. Here, we used two different machine learning models namely support vector machine and binary convolutional neural network. These algorithms gave a promising results of 79% and 89.5%, respectively. Various analytical methods are used to evaluate the performance of these models.