This repository is a Computer Vision coursework project. We aim to create an image classifier that discriminates pictures of household objects, such as mugs, boxes, and toys. The given dataset contains 50 object classes; and four images are provided for each class. These four images are taken from different angles at the same well-illuminated object, with a clean, slightly reflective background. Our final solutions are based on SIFT, color histogram, GIST, and neural network; and have achieved a classification accuracy of 98.9%.