Computer Vision Exercise Session 8 – Condensation Tracker.

13
Computer Vision Exercise Session 8 – Condensation Tracker

description

General Tracking Framework 1. Prediction, based on system model f = system transition function 2. Update, based on measurement model h = measurement function is the history of observations

Transcript of Computer Vision Exercise Session 8 – Condensation Tracker.

Page 1: Computer Vision Exercise Session 8 – Condensation Tracker.

Computer Vision

Exercise Session 8 – Condensation Tracker

Page 2: Computer Vision Exercise Session 8 – Condensation Tracker.

Assignment Tasks

1.Condensation tracker with color histogram observations

2.Experiment with the condensation tracker

Page 3: Computer Vision Exercise Session 8 – Condensation Tracker.

General Tracking Framework

1.Prediction, based on system model

f = system transition function2.Update, based on measurement

model

h = measurement function is the history of observations

),( 111 tttt wxfx

),( tttt vxhz

), ... ,( 1 tt zzZ

Page 4: Computer Vision Exercise Session 8 – Condensation Tracker.

Condensation Tracker

The probability distribution is represented by a sample set S

- weights giving the sampling probability

NnsS nn ... 1|),( )()(

Page 5: Computer Vision Exercise Session 8 – Condensation Tracker.

Condensation Tracker

1.PredictionStart with , the sample set of the previous step, and apply the system model to each sample, yielding predicted samples

2.UpdateSample from the predicted set, where samples are drawn with replacement with probability (using measurement model)

1tS

)(' nts

)|( )(')( ntt

n szp

Page 6: Computer Vision Exercise Session 8 – Condensation Tracker.

Condensation Tracker

Samples may be drawn multiple times, but noise will yield different predictions

Page 7: Computer Vision Exercise Session 8 – Condensation Tracker.

Task 2:Experiment with the Condensation Tracker

• Moving hand• Uniform

background

• Moving hand• Clutter• Occlusions

• Ball bouncing• Motion model

Page 8: Computer Vision Exercise Session 8 – Condensation Tracker.

Video 1: Hand, uniform background

a priori mean state

a posteriori mean state

Page 9: Computer Vision Exercise Session 8 – Condensation Tracker.

Video 2: Hand, clutter, occlusions

a priori mean state

a posteriori mean state

Page 10: Computer Vision Exercise Session 8 – Condensation Tracker.

Video 3: Ball bouncing

a priori mean state

a posteriori mean state

Page 11: Computer Vision Exercise Session 8 – Condensation Tracker.

Report MATLAB code

We provide the overall structure Write the code to perform each step of the

CONDENSATION tracker

Plot the trajectories of the mean state

Experiment different settings number of particles number of bins for quantization updating appearance model motion model

Try your own video (bonus)

Page 12: Computer Vision Exercise Session 8 – Condensation Tracker.

Hand-in

Hand in* by 4pm on Thursday 26th November 2015

[email protected]

*Details on exercise sheet

Page 13: Computer Vision Exercise Session 8 – Condensation Tracker.

Next Week

No exercise session next week*!

*Unless opposite is told in the following days..