Identification of plants using leaf based on convolutional neural network

Abstract

Plants are useful to humans by providing food, medicine, fuel, fibre, shelter etc. But it is important to identify the type and uses of a plant to utilize its benefits. So, we have proposed an automated deep learning algorithm to classify plants into appropriate taxonomy using a leaf. Different plant images are captured in a natural environment and created a Leaf dataset containing 12798 leaf images with white background. Preprocessing, segmentation and pattern matching techniques were used to obtain the desired output. Convolutional Neural Network has been used as a pattern matcher to compare the information of an input image with the images in the dataset. Thus, we obtained an accuracy ranging from 95% to 99% by using Convolutional Neural Network algorithm. Hence, this paper will be useful to identify plants using a leaf for Botanists, Industrialists, Food Engineers, Physicians, etc.