This project is using http://webgazer.cs.brown.edu and has been modified for our purposes.
This project requires node to be installed locally in order to run.
To prepare this project, run the following into the terminal:
git clone https://github.com/StevenSeiden/WebGazer-Node.git
cd WebGazer-Node
chmod +x run.sh
To run this project, open the folder that the project is stored in using your terminal (using the cd command) and run the following:
./run.sh
Running this script should open a window in your browser. If this does not happen, open your browser and navigate to http://localhost:3003.
First, go to https://github.com/StevenSeiden/WebGazer-Node/ and click the green "Code" button in the upper right of the screen. Then click "Download ZIP". Save this file, and when the download has completed, extract the contents of the ZIP file.
To run this project, open the folder that the project is stored in within the command prompt (using the cd command). The path should look something like this:
C:\Users\user\Downloads\WebGazer-Node-main\WebGazer-Node-main\
Finally, run the following:
node app.js
Then open your browser and navigate to http://localhost:3003.
Once the tracker is setup, the algorithm may need to be calibrated. (The default calibration is probably fine enough, however.) To do so, simply move your cursor to the area that you are gazing at and click. The more times you do this, the more accurate the tracking will be. Try looking at different areas on your screen so that the algorithm has an idea of where your entire screen is.
To begin recording data, press 1. When you are done recording data, press 2. Data is stored in the experiment folder within the project folder (.../WebGazer-Node/experiment), and categorized by date. This folder will need to be created by the user before data collection can start. Data will not be stored otherwise.
The dot on the screen of the eye tracker represents the relative position in which the user is gazing. The data that is exported from the tracker is the coordinates of this dot on the user's screen. Additionally, there is not a set sample rate, as the algorithm runs as fast as your computer can handle. I have personally found that this algorithm reports a gaze locations about 30 times per second, which is only limited by my webcam's frame rate.
There are a lot of common mistakes that can occur when trying to get this eye tracker to work properly.
If you are having trouble with recording data, first make sure that you have created a folder called experiment within the WebGazer-Node folder. If data is still not being recorded, try to refresh the page.