Usually eye-tracking devices need to be manually calibrated by the user. This one calibrates itself without the user noticing at all. For a detailed description of the underlying concept have a look at this paper. I did this project for my bachelor thesis which is part of eyeTrax: an eye tracking system for diagnosing light concussions based on occulomotoric deficits.
Create and activate the conda environment from the env_linux_64.yml file:
conda env create -f env_linux_64.yml
conda activate ba_env
NOTE: As the filename already says, it only works with Linux 64bit
Run execute.py to simulate and eye tracking session. For setting the parameters of the
simulation have a look at the documentation of execution.py.
It should output the results as a Matplotlib figure.
Alternatively to simulating, you can use your own recorded tracking data. This is an
experimental feature yet. The recorded pupils need to be provided as 2D ellipse.
All pupils are saved in a python list and each pupil is a dictionary in the following
form.
{"x_center": self.ellipse.x_center, "y_center": self.ellipse.y_center, "maj": self.ellipse.major, "min": self.ellipse.minor, "rot": self.ellipse.clockwise_rot}
The rotation is measured in degrees. You need to replace the ellipse_param_list with
your own in the executin.py file.