Bag of Words Image Classifier

Tianyi Zhang


Supervised by Ralph Martin; Moderated by Yukun Lai

The goal of this project is to implement an image classifier based on the bag of words concept. The idea is to compute a set of features in each image, and make a histogram of feature values for each feature which acts as a fingerprint for a given kind of image. Training images will be used to learn what the histograms look like for various classes of images such as human portatis, and seascapes. New images can then be presented to the system which will identify what kind of images is present.

Final Report (11/09/2015) [Zip Archive]

Publication Form