This project integrates webgazer; an eyetracking library to a simple website. A live demo is available here This projects has been built to collect gaze tracking data from participants for calculating attention score while attending an online class or during watching an online tutorial.
* Yields best performance when the participants
* sit infront of a webcam at eye level.
* keep their head as still as possible during the calibration process and keep their head in similar pose and position during tracking.
* The eye tracking can require some processing power. So number of gaze points per second will vary from device to device.
* Accuracy is lower on the edges compared to center positions. But one might be able to improve that by tweaking the `calibration_points` array or using a different approach for calibration. **Warning:** Over calibration will cause the model to overfit and might reduce performance.