NATHAN DE LA CRUZ SUPERVISOR: MEHRDAD GHAZIASGAR MENTORS: DANE BROWN AND DIEGO MUSHFIELDT Lie Detection System Using Facial Expressions.

NATHAN DE LA CRUZ

SUPERVISOR: MEHRDAD GHAZIASGAR

MENTORS: DANE BROWN AND DIEGO MUSHFIELDT

Lie Detection System Using Facial Expressions

Introduction

BackgroundResearch has found:

More than 80% of women admit to occasionally telling “harmless half truths”.

31% of people admit to lying on their CV’s.

60% of people lie at least once during a 10 minute conversation and on average tell 2 to 3 lies.

Ways to detect Lies

Study body language

Ways to detect Lies

Studying Eye movements

LIE TRUTH

Ways to detect Lies

Observing micro-expressions

User Requirements

The user requires the system to accurately tell if a person is or is not lying.

The user should be able to initialize the software.The user will ask the subject a series of questions.The software should be monitoring the subjects’

response. At any time the user should be able to stop the

processing and get a result.

Requirements Analysis

What Is Needed?

A Web Camera

A PC with Open Computer Vision (OpenCV) libraries Installed.

Requirements Analysis

Complete Analysis

Process Initialization (Clicking Some Button)Capturing Video In Real Time & Detecting The

FacePre-processing FramesProcessing Frames Using Optical FlowProcess Termination (Clicking Some Button)Displaying Information to User

Requirements Analysis

Capturing Video In Real Time & Detecting The Face

Requirements Analysis

Pre-processing FramesGreyscale Cropping

Requirements Analysis

Processing Frames Using Optical Flow

Displaying Information to User

• Either a “Passed” or a “Failed” message will be displayed

• User not faced with detailed information

• Improves the user understandability aspect of the software

Requirements Analysis

Project Plan

Goal Due date• Learn to use OpenCV functions/tools to

manipulate images and videos• Requirements Gathering

End of Term1 (Completed)

Design and Development• Creating User Interface Specification• Designing structure of code• Identifying 2 micro-expressions and 2 macro-expression

End of Term2 (Completed)

Implementation• Training SVM to identify more micro-expressions• Optimizing LBP by altering the smoothing function

End of Term3

Testing and Evaluating• Collect more training data for SVM• Collect more test data for SVM End of Term4

References

1. Robert Etherson. (2008). people admit to lying on their resumes. Available: http://www.databaserecords.com/blog/31-of-people-admit-to-lying-on-their-resumes/. Last accessed 28th March 2013.

2. Michelle Adler. (2009). little white coat lies. Available: http://www.newsweek.com/2009/01/07/little-white-coat-lies. Last accessed 28th March 2013.

3. Henry Bach . (2004). read face deciphering micro-expression. Available: http://www.divinecaroline.com/22189/83672-read-face-deciphering-microexpressions. Last accessed 28th March 2013.