Leaf App: Leaf recognition with deep convolutional neural networks

Abstract

In this paper, a very deep convolutional neural network is used to do leaf recognition. In order to predict location of leaves, some pre-processing technique is adopted to extract regions in the image before doing classification. To improve the accuracy, we enlarge the dataset by data augmentation, i.e., doing several transformations such as horizontal reflection, contrast enhancement and rotations. Experimental results show that by using deep convolutional neural network with data augmentation, our system can achieve accuracy close to the state-of-the-art systems.