Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Lie Detection from a Video Sequence

No description
by

zoha mujahid

on 15 May 2014

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Lie Detection from a Video Sequence

Lie Detection from a Video Sequence
Case Study: Job Interviews
General Overview
The objective of our project is to explore the robustness of detecting deception through facial micro-expressions.
Deception:
An art of bluffing by managing verbal and non-verbal messages.

Micro-Expressions:
Facial expressions that arise for a very short duration but can not be controlled and easily leaked.
Motivation
Need of an automated deception detection system in:
Recruitment
Business
Security
Break Down of Project
Our
Project
Research
Experimentation
Development
Features/Cues
Identification
Features/Cues
Effectiveness
Verification
Implementation
of Cues
Computer Vision
Graphical User
Interface
Research
Literature Review
Psychological Theories
Paul Ekman
Charles Darwin
DePaulo & Morris
NLP by R.Bandler & J. Grinder
Outcome
Features Identified
Eye Movement (NLP Theory)
Duping Delight
Lip Pursing
Mouth Shrug
Eye-Brow Raising
Research
Experimentation
Developing Questionnaire
Conducting Test Interviews
Analyzing Collected Data-set
Outcome
Features Effectiveness Verified
Research
Literature
Review
Cues of
Deception
Compiling
Data Set
Manual Cues
Selection for
Implementation
Development
Results
Implementation of Cues/Indicators
Tools used:
Visual Studio
A-4 Tech Web-cam 333A
APIs used:
OpenCV
Flandmark
Language used:
C++ (Image Processing Techniques)
Development
Implementation
1: Eye Movement
Face Detection
(Haar Cascade Classifier)
Eyes Detection
Haar Cascade Classifier
Pupil Detection
Harris Corner Detection
Eyes Corners Detection
Flandmark
Distance Calculation
Between Pupil & Eye Corners
Eye Movement
Identified
Development
Implementation
2: Eye Brow Raising
Face & Eyes Detection
Haar Cascade Classifiers
Eye-Brow Detection
Using ROI of Eye
Coordinates Returned by Haar Cascades
Edge Detection
Morphological Operations (Gradient operator)
Thresholding
Dilation
Erosion
Eye-Brow Mid Point Detection
Using Moments
Distance Calculation
Between standard midpoint and the actual midpoint .
Eye-Brow Raised
Development
Implementation
3: Duping Delight
Face & Mouth Detection
Haar Cascades
Mouth Edges & Corners
Edges-> Gradient Operator
Thresholding
Dilation
Erosion
Distance between
mouth corners
Duping Delight
Development
Implementation
4: Lip Pursing
Face & Mouth Detection
Haar Cascades
Mouth Edges
Morphological Operations
Thresholding
Erosion
Dilation
Upper Lip Corner
Detection
Using Mouth Edges
Lip Pursing Detected
Using upper lip corner
Development
Implementation
5:
Mouth Shrug
Face Detection
Mouth Detection
Mouth Edges
Lower Lip Corner Detection
Development
Algorithm
Capture Video
Mark Video Frames
Subject's Response Time
Process Video
Look for Indicators
Raise Corresponding Flags
Compute Probability
Percentage of
Lying
Add corresponding weight-ages of raised flags.
Weight-age to each indicator is assigned on the basis of data set collected.
Display Deception
Percentage
of
All Responses
Processing on Stored Videos
Interface
Capture
(from live web cam feed)
Mark Frames
Using Question Start/End Button
Browse Video
Process Video
Play Processed Video
Generate Report
Real Time Processing
Start Session
Start Question
End Response
Play Processed
Video
Response Deception
Analysis
Automatically Generated as soon as "Response End" is pressed.
Stop Session
Graphical User
Interface
Observations
Frequency of occurrence of micro-expressions is directly proportional to high stakes.
Every individual has particular deceiving habits.
Improvements
Reliability of system can be improved if upper body postures are incorporated.
Processing speed can be improved by replacing Haar classifiers with efficient face recognition techniques.
Weight-age of a particular indicator in probability percentage of lying should be assigned at run time for the individual under investigation.
Limitations
Poor/ Non-uniform illumination.
Controlled head movements.
Plain Background.
OpenCV and Visual Studio integration.
Conclusion
Micro-expressions are a reliable source of detecting deception in high stakes situations.
Their reliability increases to manifolds if they are combined with other indicators of deception.
Trained liars can even control micro-expressions to some extent.
Work Division
Zuha Mujahid
Documentation
Research
Development
Proposal Defense Presentation
Mid Defense Presentation
Final Defense Presentation
Idea Video
Literature Review
Questionnaire
Analysis of data set
Compiling Results
Computing Weightage for all indicators
Implementation of two indicators
GUI (browse,process & play functionality)
Real Time Processing Mode
Nida Usmani
Documentation
Research
Development
Poster for Design expo & open house
Tools to be used
Test Interviews
Results Analysis
Developed Algorithm
Implemented two indicators
GUI (Capture,Mark frames, Generate Report functionality)
Syeda Khoobru Gilani
Documentation
Research
Development
FYP Thesis
Progress Reports
Meeting Minutes
Literature Review
Questionnaire
Test Interviews
Data Set Analysis
Implementation of one of the deception indicator.
Results
Group Members:
Zuha Mujahid
Nida Usmani
Syeda Khoobru Gilani
References
[1] Thomas O. Meservy "Deception Detection through Automatic Unobtrusive Analysis of Non-verbal Behavior" 2005.
[2] B.Fasel , Juergen Luettin "Automatic facial expression analysis: a survey" 2001.
[3] Bella M. DePaulo, Brian E. Malone "Cues to Deception" 2003.
[4] Department of psychiatry, University of California " Darwin Deception and Facial Expression" 2003.
[5] Michal Ciesla, Przemyslaw Koziol "Eye Pupil Location using Webcam.
[6] Dacher Keltner, P.Ekman "Facial Expression of Emotion" 2000.
[7] Michael Spezio "Pinocchio's Pupil: Using Eye Tracking and Pupil dilation"
[8] Jeffery F. Cohn "Human Facial Expressions as
Adaptations" 2001.
[9] Mark G. Frank, Thomas Hugh Feeley "To catch a
Liar: Challenges for Research in Lie
Detection Training" 2003.
[10] Racheal Adelson "Detecting Deception"
2004.
[11] Paul Ekman "Telling Lies"
1985.
[12] Darwin "The expression of
emotions in Man and Animals"
1872.
Are you fine with taking office work to home?
Yes
No
Won't your family have
an issue with this?
Does that mean you don't prioritize your work?
Have you discussed this issue
before with your family?
If you have a deadline, how would you manage?
Sample Questions
Do you have prior experience
in this field?
Yes
No
Please elaborate when
and where?
Why did you opt
for this field?
Subject Specific Questions:
Project based
Experience Based
Extracurricular Activities/Hobbies Based
Paul Ekman
DePaulo & Morris
High Stakes
Pupil Dilation
Lip Pursing
Lesser Eye Contact

[10]
Concept of Microexpressions
Microexpressions as deception indicators
Duping Delight

[11]
Darwin
[12]
Universality of facial expressions
Leakage of expressions while deceiving
Full transcript