PLANT SPECIES CLASSIFCATION USING BOF

Automated plant species identification system could help botanists and layman for identifying plant species expeditiously. In this research, a BoF(Bag of Features) approach similar to that of deep learning is used to extract the features of the leaves and classify them hastily. 16 species are there in the dataset and in each species there are 30 images, so a total of 16 X 30, 480 images are in total. The leaf images are given as input and they undergoes different filtering techniques and segmentation, after that the image is fed into Bag of Features for feature extraction and classification. In olden days the shape, colour and the leaf vein morphometrics are used for feature extraction. Here the texture features of the leaves are used for extraction and classification. In addition to this, in this research the GLCM feature extraction method and Naive Bayes classification method is also used for comparing the result of Bag of Features approach. Using GLCM and Nave Bayes, get an accuracy of 89% in which in BoF 96% accuracy is obtained. From the research, we can conclude that the BoF can be an effective automated system for plant species identification. Keywords—BoF, GLCM, Naïve Bayes, Thresholding, Filtering, Segmentation.