A Novel CNN-KNN based Hybrid Method for Plant Classification

Abstract

Plant classification is an interesting problem in Computer Vision. Several researchers are completed to classification of plant by leaf of plant and flower of plant. After several research efforts, it has been confirmed that leaf of plant is the best and consistent source for classification of plant. However it is interesting to classify a plant through structure of leaf. Hence, it is mandatory to normalize the leaves of plant into the same size to acquire better performance. In this paper, we have proposed a hybrid method namely CNN-KNN based hybrid method. This method used on two datasets. These datasets are LeafSnap and Flavia. CNN + KNN achieved to get maximum accuracy of 98.4% and 96.5%, respectively, in LeafSnap and Flavia datasets