The Comparison Of Leaf Classification Performance Of Deep Learning Algorithms

In this article, the achievements of deep learning algorithms have been put forward by classifying plant leaves through deep learning although there are many methods used for the classification of plant leaves. Results are obtained through pre-treatment, feature extraction and classification method in classification processes made with image processing methods. There is no need for such operations in deep learning methods. In deep learning methods, the steps such as pre-processing and feature extraction are performed through Convolution Neural Networks. In this study, there are about 1900 images of 32 samples in the database used as leaf pattern. There is an average of 60 images for each image class. The images here have been increased 4 times by reflection and reversing operations and processes have been made with approximately 7600 images. Deep learning algorithms such as AlexNet, Vgg16, Vgg19, ResNet50 and GoogleNet have been used for leaf classification applications for each algorithm and their performance has been evaluated.