Cohort study on recognition of plant species using Deep Learning methods

Abstract

Plants play a vital role in each living organism’s life since it maintains the environment and provides us valuable medicine, food, fragrance etc. Knowledge of species is important for the protection of biodiversity. The identification of species of plants by a manual method by botanist is tedious work besides the complex botanical terms used by an expert are annoying for a non-expert. This may lead to the obstruction for learners interested in procuring knowledge of plant species. By applying the classification of species one can also capture crops from weed for automated weedicide. Species of plant recognition are a matter of huge significance in various areas of farming, maintenance of environmental, natural, manufactured goods and medicine invention, and other related areas. Leaf color leaves contour, shape, leaf size, flowers, texture, margins, etc. are the features of plants which can be used for classification, and however, extraction of traits from selected features is the most important status in the classification. In this paper, a review-based study is done which is based on approaches such as Machine learning algorithm, Deep Learning, Convolutional Neural Networks (CNN), etc. are compared. Various classification methods like K-nearest neighbor (KNN), Naïve baise(NB), Random forest are also studied. Mostly used datasets such as Flavia, swedish, Leafsnap, ICL with their species wise features were studied.