+ All Categories
Home > Documents > Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Date post: 15-Jan-2016
Category:
View: 217 times
Download: 0 times
Share this document with a friend
Popular Tags:
19
Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006
Transcript
Page 1: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Local Affine Feature Tracking in Films/Sitcoms

Chunhui GuCS 294-6

Final PresentationDec. 13, 2006

Page 2: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Objective

• Automatically detect and track local affine features in film/sitcom frame sequences.

– Current Dataset: Sex and the City– Why sitcom?

• Simple daily environment• Few or no special effects• Repeated scenes

Page 3: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Outline

• Preprocessing• Tracking Algorithm

– Pairwise local matching– Robust features

• Feature Matching across Shots• Results

– Feature matching vs baseline color histogram– Time complexity– When does tracking fail

Page 4: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Preprocessing

FrameExtraction

(i-1)’th shot i’th shot

ShotDetection

MSER Interest PointDetection

SIFT FeatureExtraction

Page 5: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Tracking Algorithm

• Basic: Pairwise Matching

Frame i Frame j=i+1

imf

,i im mx y

jnf

Page 6: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Tracking Algorithm

• Basic: Pairwise Matching

Frame i Frame j=i+1

imf

,i im mx y

jnf

Page 7: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Tracking Algorithm

• Basic: Pairwise Matching

Frame i Frame j=i+1

imf

,i im mx y

jnf min fd

Thresholding on both minimum distance and ratio

Page 8: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Tracking Algorithm

• Basic: Pairwise Matching

Frame i Frame j=i+1

imf

,i im mx y

jnf

Page 9: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Tracking Algorithm

• Basic: Pairwise Matching

Frame i Frame j=i+1

imf

,i im mx y

jnf

Page 10: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Tracking Algorithm

• Problem of Pairwise Matching– Sensitive to occlusion and feature misdetection

• Solutions:– Use multiple overlapping windows– Backward Matching

• Match features in current frame to features in all previous frames within the shot

• Pruning process (reduce computation time)

• Select a proportion of features that have longer tracking length as robust features

Page 11: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Shot grouping/Scene Retrieval

60

601 2, ,...rf mf x x x

56

561 2, ,...rf mf x x x

10746 10747 10772

Shot 49

10933 10934 10968

Shot 53

11393 11394 11435

Shot 56

Shot 60

11533 11534 11560

Scene 5

49

491 2, ,...rf mf x x x

53

531 2, ,...rf mf x x x

Page 12: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Inter-Shot Matching

Shot I Shot J

1 11 2, ,...I mf x x x

2 21 2, ,...I mf x x x

1 11 2, ,...J nf x x x

2 21 2, ,...J nf x x x

1 2, ,...q qJ nf x x x 1 2, ,...

p pI mf x x x

D

Page 13: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

“Confusion Table”

Ground Truth50 55 60 65 70 75

50

55

60

65

70

75

Color Histograms50 55 60 65 70 75

50

55

60

65

70

75

Feature Matching50 55 60 65 70 75

50

55

60

65

70

75

Page 14: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

ROC

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

False Alarm

Tru

e D

etec

tion

ROC curve of Feature Matching

Page 15: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

When Does Tracking Fail?• Tracking feature outside local window

– Rare when continuous tracking– Happens when occlusion occurs

• Same feature splitting to two or more groups– Long occlusion– Multiple matching in a single frame

Frame i Frame j=i+1

imf

,i im mx y

jnf

Page 16: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Computation Complexity• Everything except for MSER and SIFT algorithms are

implemented in Matlab (slow…)

Complexity Time

Frame Extraction O(N) ~0.3s/frame

Shot Detection O(N*f(B)) ~0.07s/frame (B=16)

MSER Detection O(N) ~0.3s/frame

SIFT Detection O(N) ~0.9s/frame

Feature Tracking O(N*F*W*L) ~0.5s/frame

Matching across shots

O(S2*T2) ~1s/shot pair

N: # of frames; (30,000) B: # of bins for color hist (16) F: ave. # of features per frame; (400) W: Local window size; (15)L: tracking length; (20) T: ave. # of robust trackers per shot; (300)S: # of shots; (35)

Page 17: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Conclusion

• We successfully implemented local affine feature tracking in sitcom “sex and the city”. The tracking method is robust to occlusion and feature misdetection.

• Although no quantitative precision/recall curve (hard to find ground truth), the demonstration shows that precision is almost perfect with good recall performance.

• We show one successful application of using robust features to associate similar shots together for scene retrieval.

Page 18: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Future Work

• Implement algorithm in real-time (C/C++)

• Search unique shots in films/sitcoms

• Separate indoor scenes from outdoor scenes

• Determine context of the scene

Page 19: Local Affine Feature Tracking in Films/Sitcoms Chunhui Gu CS 294-6 Final Presentation Dec. 13, 2006.

Acknowledgement


Recommended