Eyetracking get involved into virtual patient system

1
Eyetracking get involved into virtual patient
system
2
Structure

• Why?
• No real-time tracking data output for ASL
  eyetracker
• Interface between OpenCV and frame grabber
  Intel Performance Library Authorization

Data analysis
3
Eyetracking logs

• originally EYD file not readable
• Have to be converted to asc file to be read
• Key information including
• gaze coordinates of the scene
• pupil diameter

4
Eyetracking video

• Record method
• Camstudio
• Why? Frame grabber
• Error exists because of the indirect output
• The video is not originally read by openCV
• The conversion can cause quality sacrifice

5
Face detection

• The biggest issue
• The detection performance
• If the gaze line affects? No resolution
• What if use direct video input?

6
Data Analysis

• The data from two logs have to be adjusted into
  the same coordinate system.
• Gaze distribution analysis ( compared with face
  detection)
• Go down to the specific position of the video.
• The two logs have to be recorded at the same
  time, or there will be error

7
Issues Addressed

• Head gaze
• Quantitive result
• Gaze tracking video