A Performance Assessment of Rose Plant Classification Using Machine Learning

Abstract

Machine learning enriches the field of artificial intelligence that aims to make computers powerful by providing them information extracted from data. Flowers identification is highly significant and relevant for Plant Scientists. Carrying it out manually is not only a tedious task but also prone to errors due to a large number of flower types. Using machine learning algorithms to identify flowers is appealing. To this aim, two observations on flower leaves are relevant and leverage flower identification: one, flower plants have key knowledge in their leaves, thus enable distinctiveness; two, leaves have a much longer life on plants than flowers and fruits. In this paper, we have proposed a machine learning approach based on k Nearest Neighbor (k-NN) to identify rose types. Following steps are carried out during the identification process. First, rose plant images are taken using 23MP camera, ensuring temperature uniformity during the experiment. Second, texture and histogram features are extracted from the captured images. Third, k-NN algorithm is applied to these features with k taking values between 1 and 10. Our research brings to limelight the usefulness of selected features for rose type identification with histogram and texture features achieving maximum accuracies of 65% and 45.50% respectively.