Pupil-Core EyeTracker

Introduction:

• The tools under this project are used for processing and visualizing the data retrieved from the application named Pupil-Core.
• As a first step, one needs to export data from PupilCore application.
• There are two versions for visualizing the data:
  • Real-time visualization on Video (via Command Terminal)
  • Visualization on Dashboard( Jupyter Notebook)

How to get the exported files from Pupil-core:

• Make sure to activate the plugins given below:
  • Blink Detector
  • Head Pose Tracker
  • Raw Data Exporter
  • Surface Tracker
  • World Video Exporter
• Then check certain blink threshold on the Blink Detector. Check the values of 1 which suggests occurance of a blink, tighten the range till you get a subset of blinks correctly, false positives excluded.
• Then enter Head Pose Tracker, press calculate.
• Then press the Download icon on the left to export the files you need.
• You will find the files needed scattered around inside the path like 000/exports/#Number_Of_Export.

Real-time visualization on Video:

Description

• This tool processes the data exported from Pupil-core and visualizes the data in the form of:
  • Head-pose and gaze plotting on a screen : Combining both features we are estimating approximately where the user is looking at the screen
  • Blink-rate : To interpret certain moments the user is experiencing, such as excitement, anxiety, we are checking blink-rate
  • Eye-Dilation : To interpret driver's focus.
  • Heatmap of gaze: Where user gazed in the screen

How to run through Command Terminal:

$ python main.py /
            --headpose_tracker resources/000/head_pose_tracker_poses.csv /
            --marker_detections resources/000/marker_detections.csv /
            --pupil_positions resources/000/pupil_positions.csv /
            --gaze_positions resources/000/gaze_positions.csv /
            --blinks resources/000/blinks.csv /
            --video resources/000/world.mp4 /
            --world_timestamps resources/000/world_timestamps.csv

• The scripts has lots of arguments. So it would be easy to use for users to stick with the convention of putting in the same folder for their sake.

Visualization on Dashboard:

Description

• This tool also processes the data exported from Pupil-core and visualizes the data in the form of:
  • Head-pose and gaze plotting on a screen : Combining both features we are estimating approximately where the user is looking at the screen
  • Eye-Dilation : To interpret driver's focus.
  • Heatmap of gaze: Where user gazed in the screen
  • Blinks
  • Marker Counts
• But this tools, allows the cumulative graphs to be created between specified indexes. So allow the user to do comparison much easier.

How to run Visualization on Dashboard:

• Export files as descibed above

  • You will find the necessary files as described from an example below:

    • head_pose_tracker_poses.csv file would be stored in 000/exports/000/head_pose_tracker_poses.csv which should be set as headpose_tracker_poses_path

    • marker_detections.csv file would be stored in 000/exports/000/marker_detections.csv which should be set as marker_detections_path

    • world_timestamps.csv file would be stored in 000/exports/000/world_timestamps.csv which should be set as world_timestamps_path

    • pupil_positions.csv file would be stored in 000/exports/000/pupil_positions.csv which should be set as pupil_positions_path

    • gaze_positions.csv file would be stored in 000/exports/000/gaze_positions.csv which should be set as gaze_positions_path

    • blinks.csv file would be stored in 000/exports/000/blinks.csv which should be set as blinks_path

• Install jupyter

• Start jupyter lab with the following code in any terminal jupyter lab

• Put the file-paths inside the Configure Filepaths cell. As shown in the example below:

• Run the code in any jupyter lab and the code will run.