You are on page 1of 23

RECOGNIZING FACIAL EXPRESSIONS THROUGH TRACKING

Salih Burak Gokturk

OVERVIEW
MOTIVATION PREVIOUS WORK PROBLEM DESCRIPTION THEORY OF TRACKING TRACKING VIDEOS THEORY OF SVM

MOTIVATION
What does expression recognition mean?
Guessing the meaning of facial deformations.

What are the possible applications ?


- Any interactive scenerio. - Video conferencing

Why do we want to use 3-D information ?


- 2-D system is very dependent on the view yet computationally simple.

Components of the recognition system

Analysis -Face Tracking

Shape Parameters

Intelligence -Support Vector Machine Classifier

OVERVIEW
MOTIVATION PREVIOUS WORK PROBLEM DESCRIPTION THEORY OF TRACKING TRACKING VIDEOS THEORY OF SVM

SNAKE MODEL
Introduced by Kass and Witkin. Energy Minimization Problem:

Used by Waters and Terzopoulos for tracking. Snakes are fit to important regions, and tracked from one view to the other.

MODEL BASED 3-D FACE TRACKING


DeCarlo and Metaxas, 96, very accurate tracking. Eisert and Girod, 98, used in video-conferencing application, with efficient and accurate compression. Gokturk et. al., 00, brings a data driven approach where the face model is learnt from stereo tracking. Model based approaches. Track all the points together with n-dimensional freedom on the shape. Based on Lukas-Tomasi-Kanade optical flow tracker.

OVERVIEW
MOTIVATION PREVIOUS WORK PROBLEM DESCRIPTION THEORY OF TRACKING TRACKING VIDEOS THEORY OF SVM

PROBLEM DESCRIPTION(Tracking 1)

PROBLEM DESCRIPTION(Tracking 2)
X(t) X(t+1)
TIME t+1

?
I(t+1)

I(x(t))

PROBLEM DESCRIPTION (Recognition)


Data Training Classifier New Data Testing Output

X(t)

[ Rigid, Open Mouth, Smile]

?
[ Rigid, Open Mouth, Smile]

OVERVIEW
MOTIVATION PREVIOUS WORK PROBLEM DESCRIPTION THEORY OF TRACKING TRACKING VIDEOS THEORY OF SVM

ASSUMPTIONS
Cameras are calibrated. The person should move slow unless the camera is fast enough for motion capture. The mesh is initialized to the first image. The user performs the expressions known to the computer

Data

Stereo Tracking

Learn Shape

Monocular Tracking

- shape is learnt from stereo learning in our case

X = X0 + i Xi
i =1

p - degrees of freedom

LUKAS TOMASI KANADE OPTICAL FLOW TRACKER


Time t:
Il(xi(t))

u I x I y = I t v

?
Time t+1:
Il(xi(t+1))

- For robustness, u and v are estimated using a neighbourhood around the point.

LUKAS TOMASI KANADE OPTICAL FLOW TRACKER EXTENDED TO 3D


X(t) X(t+1)

X = X0 + i Xi
i =1

u I x I y = I t v

X ( R, T , )

TIME t+1

I(x(t))

I(t+1)

u u dR u u dR dT d dT v = v v v d dR dT d
J

[I

Iy

dR J dT = I t d

OVERVIEW
MOTIVATION PREVIOUS WORK PROBLEM DESCRIPTION
THEORY OF TRACKING TRACKING VIDEOS THEORY OF SVM

STEREO LEARNING

FACE TRACKING
- The deformation space of a particular individual is learnt - That particular individual is tracked using a mono camera
X = X 0 + i X i
i =1 P

- Tracked parameters : , R, T.

-The deformation space of a subset of people learnt - The shapes are aligned and PCA is applied on this set

UNIVERSAL FACE TRACKER

OVERVIEW
MOTIVATION PREVIOUS WORK PROBLEM DESCRIPTION
THEORY OF TRACKING TRACKING VIDEOS THEORY OF SVM

Support Vector Machines (SVM)


Data Training Classifier New Data Testing Output

- Best discriminating hyperplane between two class of objects - Distinguish the vectors that carry the relevant information (support vectors) - if nonlinear data, map the data to high dimensional domain, then apply the SVM.

What needs to be done


- Choose the appropriate data input for SVM classifier. Using vector might not help. Create more intelligent vectors in that case. - Choose an appropriate kernel (transformation function) for SVM. - Apply SVM in a one to many fashion. - Combination of intelligent features and SVM should give robust results.

Expected Contributions
- Show that 3-D model based tracking is suitable for further applications. - Support vector machine is a suitable classifier for expression recognition

You might also like