Classification of Succulent Plant Using Convolutional Neural Network

Abstract

Succulent plant-like aloes have many medical uses. They use as a laxative, to treat joint pain, skin inflammation, conjunctivitis, hypertension, stress, etc. It keeps releasing oxygen all night where other plants release carbon dioxide at night. Classification of succulent plant is a challenging and important topic to solve for their complex shape and beauty. Deep learning approach has a strong ability to extract high-level features from a piece of an image. This paper will introduce a new dataset of succulent plant and all data of the dataset are real data. We apply deep convolutional neural networks (DCNNs) on our dataset. First of all, we use the open camera for data collection and the camera resolution is 640×480 in order that the size of our image is same. Then we collect all the data from different nurseries in Dhaka city. We choose 9 different classes of succulent plant. Our total image 3421 where training data set is 2612 and the test data set is 809. We have used three convolutional layers, three max-pooling layer, three dropout layer and we also have a fully connected layer. We have run 30 epoch in our model and our result is really good. We achieve 97.40% accuracy using our model which is the best of all previous.