Overview

This repository contains the files, work and final report for the NYU Foundations of Machine Learning graduate class for Fall 2016.

We sought to expand on previous work in gaze tracking by testing traditional regression models against the collected data set consisting of 50000 image screen-coordinate pairs.

With respect to the folders,

• datacollection contains the application built in Objective-C used for datacollection
• preprocess contains the OpenCV application built to preprocess the images
• report.pdf is the final report written for the class
• Moreover pipeline.py was used to train the models, while data-final50000.txt contains the final data. Lastly, the matlab file contains the image saliency map production. Note that you will need SaliencyToolbox (couple of Mb, download at www.saliencytoolbox.net/) by Walter.

In order to run the models run

python pipeline.py data-final50000.txt 1 1
python pipeline.py data-final50000.txt 0 1

in the terminal.

Any feedback is greatly appreciated.

Frederik Jensen and Jovan Jovancevic