Neural Eye Tracking for Human Computer Interaction
Presented Presented By: Sibarama Panigrahi(11040056) Sibarama Panigrahi(11040056) VSSUT,Burla VSSUT,Burla
Contents
Input Frame
Frame Search Region
Pre-Processing
Bayesian Classifier
Clustering
Post-Processing
Yes Eyes Located Successfully?
Update frame Search Region
Update Means And Covariance.
Yes
Facts About Human Gaze
Ones eye is rarely stationary Eye movement reflects a viewers visual information process The eye can move faster than the hand Eye movement consists of:
Saccades Fixations
Why Use Eye Tracking?
Gives an accurate measure of where ones visual attention is directed Enhances or backs-up observations May reveal patterns which may become proof of effective or ineffective usability Can lead to many potentially useful applications
Types of Eye Tracking
Skin Electrodes Contact Lens Head Mounted Remote Systems
(Desk or Computer-based)
Types of Eye Tracking:
Skin Electrodes
Electrodes placed on the skin around the eye socket Measures electrical differences the between the retina and cornea
GOOD POINTS: Both eyes can be recorded together Least expensive Simple to use DOWNFALLS: It is limited to horizontal and vertical movements Poor accuracy for absolute positioning
Types of Eye Tracking:
Contact Lens
A non-slipping contact lens fits over corneal bulge. Tracking is recorded by affixing a magnetic coil or mirror to the lens. GOOD POINTS:
Provides
accurate data about the nature of human eye movements.
DOWNFALLS:
Extremely
awkward, uncomfortable for the user Interferes with blinking Covers only a limited range of eye movements
Types of Eye Tracking:
Head Mounted
Small camera and light source mounted to users head via a headband or helmet Reports the angle of the users eye with respect to his or her head. Two data sources can determine the line of gaze in physical space.
GOOD POINTS: Doesnt restrict the users head movements (including locations away from the computer screen) DOWNFALLS: More awkward to use than the desk-based system as the user has to have instrument mounted to head.
10
Types of Eye Tracking:
Remote Systems
Most practical method of eye tracking Uses Illuminator/eye camera Tracking visible features of the eye Head movements can be distinguished from eye movements by tracking 2 points
GOOD POINTS: Allows for a fair range of head movements Accurate, fast and affordable Not intrusive DOWNFALLS: Head still needs to stay within camera range. Delicate to calibrate and operate
11
Processing Raw Data
Data from eye tracking is too low level and voluminous for immediate interpretation. Software packages can combine data from eye tracking and interaction logs to produce a page-by-page reports.
12
Pitfall of Using Eye Tracking Technologies
Equipment is still quite intrusive
Noisy Restricted head movements Requires more patience by user
Fairly large pool of test subjects is needed
User vision must be considered
How can we tell if and what info. is really being processed?
eye fixations provide measurement of visual interest but do not provide a measure of cognitive interest.
Often viewed as a time-consuming methodology returning questionable benefits.
13
Previous Research/Applications
2 Main Categories
Interface
Usability Studies Applications
Interactive
14
Interface Usability Studies
Overview
Challenging of the controversial Scanpath Theory (Noton and Stark, 1971) The Scanpath Theory
Argues that its improbable that the eye is controlled by specific features of a visual display Rejects the explanation that subjects are driven by habits Since Internet users are repeatedly exposed to certain visual displays of information, the Web is an ideal stimulus to test this theory. Many studies dispute the Scanpath Theory
15
Interface Usability Studies
Eye Tracking CAN
Tell whether users are looking at the screen Tell whether users are reading or scanning. Learn the relative intensity of a users attention to various parts of a web page
Determine whether a user is searching for a specific item Compare user scan patterns
16
Interface Usability Studies
Eye Tracking CANT
Let you know whether users actually see something Prove that users didnt see something Determine why users are looking at something Test everybody
17
Interface Usability Studies
Online News portal study done by Stanford University and The Poynter Institute in 2000
18
Research Applications/Studies:
Online News Portals
by Stanford University and The Poynter Institute (2000)
the first eyetracking study documenting the actual second-by-second behavior of users of online news sites
Tracked what subjects read and their movements from page to page.
Used 67 subjects in two cities Subjects were readers of online news - people who visited newsrelated sites at least three times a week.
Each screen was analyzed by: Format (what layout elements attracted users) Topics Actions Time spent on page
19
Research Applications/Studies:
Online News Portals
The numbers superimposed over the pink blocks show the order in which fixation clusters. Follow the numbers and note the roaming pattern. The dark "x" to the right of photo that indicates a mouse click for calling up a hyperlinked story.
20
Research Applications/Studies:
Online News Portals
Return to main page after link click
21
Research Applications/Studies:
Online News Portals
OVERALL CONCLUSIONS:
A strong preference for text over graphics (briefs or captions)
The eyes of online news readers then come back to the photos and graphics
Banner ads do catch online readers' attention
for an average of one second. Long enough to perceive the ad (but what about animated ads?)
Statistics of how different users (by age, sex) navigate through online news
22
Interactive Applications
Accessibility System Enhancement Non-Command Based Systems Virtual Displays
23
Interactive Applications:
Accessibility
Eye tracking can allow people to use their eyes to communicate Its users include people with brain injuries, cerebral palsy, multiple sclerosis, muscular dystrophy, spinal cord injuries, spinal muscular atrophy, strokes, etc.
24
Interactive Applications:
System Enhancement
Aimed toward users whose hands are occupied Potential Problems:
Some will find the eye movement based interface better (faster, more convenient and more natural) while others may feel uncomfortable. Too unnatural to use in critical situations?
Eye tracking may be best used to act as a supplemental input or display method.
25
Interactive Applications:
Non-Command Based Systems
The computer passively monitors the user and responds as appropriate, rather than waiting for the user to issue specific commands.
Eye-controlled
paddleball game
The
Little Prince (interactive childrens story)
The user controls the flow of the narration by where he or she looks without explicit commands (interactive fiction)
Problem: How can we issue a command clearly?
Gaze
duration and blnking can be ineffective or faulty
26
Interactive Applications:
Virtual Displays
Making VR more realistic
If accurate, the user cannot detect the difference between this arrangement and the large, high-resolution display it emulates.
27
Advantages of Using Eye Tracking as an Input Device
Eye movement is faster than other current input media No training or particular coordination is required of normal users Can determine where the users interest is focused automatically
28
Disadvantages of Using Eye Tracking as an Input Device
People are not accustomed to operating devices by simply moving their eyes. In interactive applications, using the eye as an input device, when does the system decide that a user is interested in a visual object?
29
Future of Eye tracking?
Helpful for usability studies to understand users interact with their environments. Potentially could provide new and more effective methods of computer-human interaction.
Eye tracking equipment is still improving, and is not quite there yet but has an exciting future!
30
VIDEOS
31
VIDEOS
32
VIDEOS
33
Links:
RESEARCH LINKS
Gaze-Based Interaction Group [Link] A page of links about various eye tracking projects and research studies User Interface Engineering [Link] Stanford Poynter Project [Link] Critique on Study: [Link] Technology News: New Software Replaces Computer Keyboards with Eyes: August 23, 2002 [Link] An interesting article on a software has been named "Dasher", that claims to "exploit our eyes' natural ability to navigate and identify familiar patterns. It is the result of work done by a team at Cavendish Laboratory in Cambridge, UK. An eye-tracking device allows users to select letters from the screen. The Little Prince (interactive childrens story experiment) I. Starker and R.A. Bolt, A Gaze-Responsive Self-Disclosing Display, Proc. ACM CHI 90 Human Factors in Computing Systems Conference pp. 3-9, Addison-Wesley/ACM Press (1990).
34
Links :
ANALYSIS SOFTWARE
SensoMotoric Instruments [Link] specializes in the development and integration of eye movement measuring and evaluation systems Seeing Machines [Link] Creators of faceLAB -a head, face and gaze tracking technology that measures the position of a human head, gaze direction and blink events. GazeTracker [Link] Software that consolidates the information that researchers need to conduct their web design and software usability studies into one piece of software. EYETECH DIGITAL SYSTEMS, INC. [Link] Software to operate your PC hands free ARRINGTON RESEARCH, INC. [Link] -creators of eye tracking hardware and software
35
Links :
EYE TRACKING EQUIPMENT
Eye Movement Equipment Database (EMED) [Link] Eye Movement Equipment Database, based in the Institute of Behavioural Sciences, University of Derby. LC Technologies [Link] LC Technologies designs, manufactures, and distributes unobtrusive video eyetracking systems. Metrovision [Link] A company located in France that sells instruments for ophthalmologists including visual field, visual electrophysiology, contrast sensitivity, and eye movements. Applied Science Laboratories [Link] develops eye tracking technology and systems SR Research [Link] Creators of EyeLink II an eye tracking system that consists of three miniature cameras mounted on a comfortable leather-padded.
36
Bibliography
Crowe, Eric C. and Narayanan, Hari. Comparing interfaces based on what users watch and do. In Proceedings of Eye Tracking Research & Applications Symposium 2000, pages 29 -- 36. ACM, 2000. Hayhoe, Mary M. , Ballard,Dana H., Triesch, Jochen, Shinoda, Hiroyuki, Aivar, Pilar, and Sullivan, Brian. Vision in Natural and Virtual Environments. In Proceedings of Eye Tracking Research & Applications Symposium 2002, pages 7 -- 13. ACM, 2002.
Jacob. Robert J.K., "Eye Tracking in Advanced Interface Design," HCI Lab, Naval Research Laboratory [Link]
Josephson, Sheree (Weber State University, USA) and Holmes, Michael E. (Ball State University). Visual Attention to Repeated Internet Images: Testing the Scanpath Theory on the World Wide Web. In Proceedings of Eye Tracking Research & Applications Symposium 2002, pages 43 -- 49. ACM, 2002.
Description of a study that challenges the Scanpath Theory, concluding that other forces such as visual features or memory may be of importance.
Lewenstein, M., Edwards, G., Tatar, D., and DeVigal, A. Stanford-Poynter Project. See [Link] 2000.
Schroeder, W. (1998, September/October). What Is Eye Tracking Good for? User Interface Engineering. [On-Line] Available URL: [Link]
Schroeder, W. (1998, September/October). Testing web sites with eye-tracking. User Interface Engineering. [On-Line] Available URL: [Link]
Vertegall, Roel. Designing Attentive Interfaces. In Proceedings of Eye Tracking Research & Applications Symposium 2002, pages 23 -- 30. ACM, 2002.
37