Supervised and unsupervised learning in animal classification

In this work, we have developed a supervised and unsupervised based classification system to classify the animals. Initially, the animal images are segmented using maximal region merging segmentation algorithm. The Gabor features are extracted from segmented images. Further, the extracted features are reduced based on supervised and unsupervised methods. In supervised method, we have used Linear Discriminate Analysis (LDA) dimension reduction technique to reduce the features. The reduced features are fed into symbolic classifier for the purpose of classification. In unsupervised method, we have used Principle component analysis (PCA) dimension reduction technique to reduce the features. The reduced features are fed into K-means algorithm for the purpose of grouping. Experimentation has been conducted on a dataset of 2000 animal images consisting of 20 different categories of animals with varying percentages of training samples. From the proposed model, it is observed that supervised classification system performs better compared to unsupervised method.