An Extensive Study on Machine Learning Paradigms Towards Medicinal Plant Classification on Potential of Medicinal Properties

The automatic classification of medicinal plants requires more exploration as it is considered as major issue for conservation, authentication, and manufacturing of medicines. Generally, medicinal plants have been classified by features of the leaf with respect to color, shape and texture. Leaf is a main parameter on analyzing its plant nutrition, plant contentions, plant soil-water association, plant preservation measures, crop ecosystems, plant respiration rate, plant transpiration rate and plant photosynthesis. Classification of the plant species is a primary and highly essential procedure for plant conservation. An object recognition system is required to classify the various species of the plant species and to protect them from various diseases. In this article, a detailed survey on machine learning models has been carried out to identify and classify medicinal plants by considering the texture and shape features of a plant leaf using linear and non linear feature descriptors. However the extracted features from the plant leaf image will be huge containing high redundancy information's. On employment of feature selection techniques through weighted average strategies through metaheuristic techniques, those techniques reduces the redundancy on feature extracted and minimizes the equal error rate to obtain the optimum weighted features. Further numerous classification techniques on supervised and unsupervised types has been employed to classify the optimal feature on various dataset has been experimented and validated using cross fold validation using confusion matrix. It is vital and essential task for providing detailed insight on that classification model for medicinal plant with respect to its medicinal properties. The efficacy of each model has been demonstrated on single plant and multiple plants on basis of classifier and dataset employed. Finally outline of the proposed methodology as framework to classify the medicinal plant has been provided. Evaluation of models has been carried out on the processing of the dataset.