Multi-Organ Plant Classification Using Deep Learning

Abstract

The variability in the shape and appearance of the same plant organs and similarity between organs of different plants results in fewer inter-class and high intra-class variations making organ-based plant classification a challenging problem. Classification of plants using a single organ may not be able to deal with these challenges. Thus the use of multiple organs can be more effective in improving the classification performance by learning different aspects of the same class. Existing approaches mainly focus on generic features of plants while ignoring features related to multiple organs. In the proposed approach, Convolutional Neural Network (CNN) is used to exploit the information of multiple organs instead of a single organ for the classification of plants. Moreover, the representation of minority classes is increased through DC GAN. The comparison of the proposed approach with the existing approaches on the publicly available PlantCLEF dataset shows its better performance in the accurate classification of plants.