Leaf Based Trees Identification Using Convolutional Neural Network

There are huge numbers of species of trees present on the earth. Only those who study plants can easily identify them but for a common people it’s very difficult to distinguish among them. Machine learning is an important technology of all time. Deep learning is basically a subset of a machine learning that automatically extracts best features, learns it and provides best results. In this paper we have proposed one of the deep learning’s approach model Convolutional Neural Network (CNN) that automatically classify different trees (Mango, Neem, Hibiscus, Guava, Lemon, Ashoka, Curry, Sheesham, Cachia, Henna) on the basis of their leaves and achieved the test accuracy of 99.40%. Experimental results were performed on our own local dataset of leaves, which consists of 10,000 images of leaves, one thousand for each class of trees. Results show that the proposed system provides an effective classification model for the identification of trees on the basis of their leaves.