Skip to content

Latest commit

 

History

History
81 lines (59 loc) · 2.83 KB

README.md

File metadata and controls

81 lines (59 loc) · 2.83 KB
Raw

JGaze

© 2014 Jose Manuel Gómez Poveda.

Eye tracking and gaze library in Java, offering the following functionalities:

  • Tracking user features, such as the head, eyes, pupils and eye corners, nose, and mouth
  • Blink detection
  • Generation of gaze heat maps and visualisations of saccades and fixations
  • Supports exporting data for offline analysis
  • Abstraction layer for working with video files, images or live video sources in the same way
  • WebSockets communication if the Webbit library is included

Using OpenCV for many low-level operations, so the speed is very close to that of implementations in C++.

Execution of JGaze UserMonitor

Click on the image below for launching the UserMonitor application:

alt text

Note that the sample above needs a webcam in order to capture video.

Also, given that it is not signed, the application will only run if the URL "http://jmgomezpoveda.github.io" is whitelisted in the security settings of the Java control panel.

Sample program

Sample program that performs face, eye and pupil detection in just a few lines of code:

import org.dreamcoder.jgaze.detectors.FacialFeaturesDetector;
import org.dreamcoder.jgaze.io.SimpleFrameGrabber;
import org.dreamcoder.jgaze.ui.CanvasFrame;
import org.dreamcoder.jgaze.util.OpenCvUtils;
import org.opencv.core.Mat;

public class PupilDetectMin
{
    public static void main(String[] args) throws Exception
    {
        OpenCvUtils.LoadLibrariesDialog();

        SimpleFrameGrabber grabber = new SimpleFrameGrabber(0, 640, 480, 30);

        try {
            Mat origImg = grabber.grab(false);

            FacialFeaturesDetector detector = new FacialFeaturesDetector(origImg.width(), origImg.height()).addPupilDetector();

            CanvasFrame frame = new CanvasFrame("PupilDetectMin", origImg.width(), origImg.height());

            while (frame.isVisible())
            {
                origImg = grabber.grab();

                detector.detect(origImg);

                detector.drawBoundingBoxes(origImg);
                detector.drawEyes(origImg);

                frame.showImage(origImg);
            }

            grabber.stop();
            detector.release();
            frame.dispose();
        }
        catch (Exception e)
        {
            System.out.println("Exception: " + e);
        }
    }
}

Publications

Jose Gómez-Poveda, Elena Gaudioso, Evaluation of temporal stability of eye tracking algorithms using webcams, Expert Systems with Applications, Available online 22 July 2016, ISSN 0957-4174, http://dx.doi.org/10.1016/j.eswa.2016.07.029. Full text available at http://authors.elsevier.com/a/1TRl83PiGT02IH