Deep Learning Model for Plant Species Classification Using Leaf Vein Features

Abstract

Leaf veins are one of the most important and complicated aspects of a leaf that are commonly used for plant species categorization and identification. Each plant species leaves have distinct qualitative characteristics that aid in classifying them. These extracted features help a botanist to identify the key characteristics of plants from their leaf images more correctly. The main phases included in proposed methodology are image preprocessing, feature extraction, and classification. The leaf images were initially pre-processed to make them compatible with the deep learning model. The features are condensed using bottleneck features, and the vein patterns in the leaf are identified using the Canny edge detection method and gathered features with the aid of a feature extraction model. VGG16 is a Convolutional Neural Network Model (CNN) that is identified to train and categorize the dataset. The experiment was conducted on the flavia dataset that were being gathered through the online source kaggle, which had 15 image classes. The model's accuracy was found to be 95 percent.