The present work proposes another path for classification of plant species from digital leaf images. Plant leaves can have an assortment of unmistakable elements like green and non-green hue, simple and compound shape and distinctive vein designed surfaces, a solitary arrangement of elements may not be sufficiently adequate for a viable classification of heterogeneous plant sorts. A hierarchical architectural design is proposed where numerous components are joined together for a more powerful and strong classification of the visual data. The study likewise incorporates the arrangements of customization of the feature extraction modules and classifiers for best execution. The database itself is sectioned in light of conspicuous components by visual discriminators, as this enhances proficiency. As new layers can be added to the current system to take into account up to this point obscure leaves with new qualities, the design likewise provides options of adaptability. Another Feature based Shape Selection Template (FSST) is proposed for the choice of shape features for various sorts of leaves. Broad examinations are completed on two openly accessible databases including green, non-green, simple and compound leaves with variations in shape, size and designs about exhibit the advantages of the proposed strategy over best in class procedures. © 2018 The Authors