Image classification using appearance based features

摘要

In this paper, a small set of features based on local appearance and texture is applied to the task of image recognition and classification. These features are used to train and subsequently test three different machine learning techniques, namely k-Nearest Neighbors (K-NN), Support Vector Machines (SVM) and Ensemble Learning (Bagging). A case study on a publicly available object classification dataset was conductor from which it was concluded that, while simple, the proposed approach was able to produce extremely high classification accuracies.