Transcript Slide 1

Use of Eye Movement Gestures
for Web Browsing
Kevin Juang
Frank Jasen
Akshay Katrekar
Joe Ahn
Motivation

Eye Gestures



Gaze input
Mouse gestures
Web Browsing


Common task in modern computing
Well suited to eye movement gestures


Well learned, emphasizing speed
Natural mapping of navigational directions
Prior Work

Gaze Input Issues (Ashmore)





Eye tracker accuracy
Sensor lag
Fixation jitter
Midas touch
Mouse Gesture Issues (Moyle & Cockburn)



Complicated non-linear gestures
Not truly modeless
Greater need for feedback
Prior Work

Off-Screen Targets (Isokoski)



Addressed previous gaze input issues
Eye typing maybe not a good domain
Need for physical off-screen targets
Methodology (Apparatus)





Tobii ET-1750 from Tobii Technology
17-inch TFT 1280x1024 Monitor
.5˚ accuracy with a 50 Hz sampling rate
25-35 ms Latency
28˚ visual angle vertically and horizontally
Methodology (Apparatus)
Methodology (Stimulus)
Methodology (Implementation)






Gaze point determined every 20 ms
Four hot zones
Middle 30% of each edge
Extended 15% off the screen
100 ms delay
Flags set when in hot zone
Methodology (Implementation)
Methodology (Implementation)
Hypotheses


Eye gesture should be both faster and less
accurate compared to mouse
Eye fatigue should cause eye gesture to
become slower as the number of
instructions increases.
Results
Number of subjects = 15
Results
Discussion

Trial Length and Input Method Interaction



Suggests mouse input in certain situations
Test results perhaps not indicative of real world
Speed-Accuracy Tradeoff



Faster even accounting for error correction
Speed benefit outweighs accuracy loss
Future results could be even more favorable
Survey Results




Ease of use (eye gestures) ---- 3.53
Speed and efficiency in using eye gestures
compared to mouse input ---- 4.53
Overall satisfaction in using eye gestures
compared to mouse ---- 3.60
Eye gesture as an effective means of web
browsing navigation ---- 3.93
Further Work

Real World Setting




High fidelity
Browser implementation
Other domains outside web browsing
Additional gestures



Corners
Following a link
Magnitude of scrolling
Conclusions

Eye Gesture System




Leverages gaze input and mouse gestures
Web browsing as a good domain
Off-screen hot zones
Benefits


Speed
User satisfaction
Questions?