recognizing human figures and actions greg mori simon fraser university

48
Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Upload: warren-fleming

Post on 12-Jan-2016

218 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Recognizing Human Figures and Actions

Greg MoriSimon Fraser University

Page 2: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Goal

• Action recognition– Where are the people?– What are they doing?

• Applications– Image understanding, image retrieval and search– HCI– Surveillance– Computer Graphics

Page 3: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

• 3-pixel man

• Blob tracking

• 300-pixel man

• Find and track limbs

Far field

Near field

Medium field• 30-pixel man

• Coarse-level actions

Page 4: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Outline

• Human figures in motion– Action Recognition

• Localizing joint positions– Exemplar-based approach– Parts-based approach

• Motion Synthesis– Novel graphics application

Page 5: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Appearance vs. Motion

Jackson PollockNumber 21 (detail)

QuickTime™ and a decompressorare needed to see this picture.

Page 6: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Action Recognition

• Recognize human actions at a distance– Low resolution, noisy data– Moving camera, occlusions– Wide range of actions (including non-periodic)

QuickTime™ and a decompressor

are needed to see this picture.

Page 7: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Our Approach

• Motion-based approach– Classify a novel motion by finding the most similar

motion from the training set– Use large amounts of data (“non-parametric”)

• Related Work– Periodicity analysis

• Polana & Nelson; Seitz & Dyer; Bobick et al; Cutler & Davis; Collins et al.

– Model-free • Temporal Templates [Bobick & Davis]

• Orientation histograms [Freeman et al; Zelnik & Irani]

• Using MoCap data [Zhao & Nevatia, Ramanan & Forsyth]

Page 8: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Gathering action data

• Tracking – Simple correlation-based tracker

QuickTime™ and a decompressor

are needed to see this picture.

QuickTime™ and a decompressor

are needed to see this picture.

Page 9: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Figure-centric Representation

• Stabilized spatio-temporal volume– No translation information– All motion caused by person’s

limbs• Good news: indifferent to camera

motion• Bad news: hard!

• Good test to see if actions, not just translation, are being captured

Page 10: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

input sequence

Remembrance of Things Past

• “Explain” novel motion sequence by matching to previously seen video clips– For each frame, match based on some temporal

extent

Challenge: how to compare motions?

run

walk leftswing

walk rightjog

database

Page 11: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

How to describe motion?

• Appearance – Not preserved across different clothing

• Gradients (spatial, temporal)– same (e.g. contrast reversal)

• Edges– Unreliable at this scale

• Optical flow– Explicitly encodes motion

– Least affected by appearance

– …but too noisy

Page 12: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Spatial Motion Descriptor

Image frame Optical flow

Fx,y

yx FF , yyxx FFFF ,,, blurred

yyxx FFFF ,,,

Page 13: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Spatio-temporal Motion Descriptor

t

Sequence A

Sequence B

Temporal window w

Bframe-to-frame

similarity matrix

A

motion-to-motionsimilarity matrix

A

B

I matrix

w

w

blurry I

w

w

Page 14: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Soccer

• Real actions, moving camera, poor video

• 8 classes of actions

• 4500 frames of labeled data

• 1-nearest-neighbor classifier

Page 15: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Classifying Ballet Actions16 Actions; 24800 total frames; 51-frame motion descriptor. Men used to classify women and vice versa.

Page 16: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Classifying Tennis Actions

6 actions; 4600 frames; 7-frame motion descriptorWoman player used as training, man as testing.

Page 17: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Classifying Tennis

• Red bars show classification results

QuickTime™ and a decompressor

are needed to see this picture.

Page 18: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Outline

• Human figures in motion– Action Recognition

• Localizing joint positions– Exemplar-based approach– Parts-based approach

• Motion Synthesis– Novel graphics application

Page 19: Recognizing Human Figures and Actions Greg Mori Simon Fraser University
Page 20: Recognizing Human Figures and Actions Greg Mori Simon Fraser University
Page 21: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Human Figures in Still Images

• Detection of humans is possible for stereotypical poses– Standing– Walking– (Viola et al., Poggio et al.)

• But we want to do more– Wider variety of poses– Localize joint positions

Page 22: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Problem

Page 23: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Shape Matching For Finding People

Database of Exemplars

Page 24: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Shape Contexts• Deformable template approach

– Shapes represented as a collection of edge points

• Two stages– Fast pruning

• Quick tests to construct a shortlist of candidate objects

• Database of known objects could be large

– Detailed matching• Perform computationally expensive comparisons on

only the few shapes in the shortlist

• Publications– Mori et al., CVPR 2001

– Mori and Malik, CVPR 2003• Featured in New York Times Science section

Page 25: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Results: Tracking by Repeated Finding

QuickTime™ and aCinepak decompressor

are needed to see this picture.

Page 26: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Multiple Exemplars

• Parts-based approach– Use a combination of keypoints or

limbs from different exemplars– Reduces the number of exemplars needed

• Compute a matching cost for each limb from every exemplar

• Compute pairwise “consistency” costs for neighbouring limbs

• Use dynamic programming to find best K configurations

Page 27: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Combining Exemplars

Page 28: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Finding People (II): Parts-based Approach

• Bottom-up

• Segmentation as preprocessing

• Detect half-limbs and torsos

• Assemble partial configurations– Prune using global constraints

• Extend partial configurations to full human figures

Page 29: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Segmentation for Recognition

• Window-scanning (e.g. face detection)– O(N M S)

SUPERPIXELS

SEGMENTS

• Segmentation– Support masks for

computation of

features

– Efficiency

– Scalability

– 600K pixels 300 superpixels, 50 segments

– O(N) + O(log(M))

Page 30: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Limb/Torso Detectors• Learn limb and torso

detectors from hand-labeled data

• Cues:– Contour

• Average edge strength on boundary

– Shape• Similarity to rectangle

– Shading• x,y gradients, blurred

– Focus• Ratio of high to low frequency

energies

Page 31: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Assembling Partial Configurations

• Combinatorial search over sets of limbs and torsos– 3 half-limbs plus a torso

configurations

• Prune using global constraints– Proximity– Relative widths– Maximum lengths– Symmetry in colour

• Complete half-limbs– 2 or 3-limbed people

• Sort partial configurations– Use limb, torso, and segmentation scores

• Extend final limbs of best configurations

Page 32: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Results

Page 33: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Results

Rank 3

Rank 3

Page 34: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Outline

• Human figures in motion– Action Recognition

• Localizing joint positions– Exemplar-based approach– Parts-based approach

• Motion Synthesis– Novel graphics application

Page 35: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

“Do as I Do” Motion Synthesis

• Matching two things:– Motion similarity across sequences– Appearance similarity within sequence

• Dynamic Programming

input sequence

synthetic sequence

Page 36: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Smoothness for Synthesis

• is similarity between input and target frames

• is appearance similarity within target frames

• For input frames {i}, find best target frames { } by maximizing following cost function:

• Optimize using dynamic programming: – N frames in input sequence– M target frames in database

Page 37: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

“Do as I Do” SynthesisTarget Frames Input Sequence

Result

3400 Frames

QuickTime™ and a decompressor

are needed to see this picture.

Page 38: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

“Do as I Say” Synthesis

• Synthesize given action labels– e.g. video game control

run walk left swing walk right jog

synthetic sequence

run

walk leftswing

walk rightjog

Page 39: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

“Do as I Say”

• Red box shows when constraint is applied

QuickTime™ and a decompressor

are needed to see this picture.

Page 40: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Frame 9½

Putting It All Together

• Can we do a better job of splicing clips together?

Frame 9 Frame 10

YES… if we can find the joints!

Page 41: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Morphed Transitions

QuickTime™ and a decompressor

are needed to see this picture.

Page 42: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

8 Transitions

QuickTime™ and a decompressor

are needed to see this picture.

Page 43: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Morphed Transitions

QuickTime™ and a decompressor

are needed to see this picture.

Page 44: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

3 Transitions

QuickTime™ and a decompressor

are needed to see this picture.

Page 45: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Actor Replacement

• Rendering new character into existing footage

• Algorithm– Track original character– Find matches from new character– Erase original character– Render in new character

• Need to worry about occlusions

Page 46: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Show the impressive video

Page 47: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Future Directions

• Much remains to be done!

• Action Recognition– Using joint positions, shape: the “morpho-kinetics” of

action recognition– Better models of activities

• Detecting and localizing figures– Combining top-down exemplar methods with bottom-up

segmentation methods– Exploiting temporal cues

Page 48: Recognizing Human Figures and Actions Greg Mori Simon Fraser University

Acknowledgements

• References– Mori, Belongie, and Malik, “Shape Contexts Enable Efficient Retrieval

of Similar Shapes”, CVPR 2001– Mori and Malik, “Estimating Human Body Configurations using Shape

Context Matching”, ECCV 2002– Efros, Berg, Mori, and Malik, “Recognizing Action at A Distance”

ICCV 2003– Mori and Malik, “Recognizing Objects in Adversarial Clutter: Breaking

a Visual CAPTCHA”, CVPR 2003– Mori, Ren, Efros, Malik, “Recovering Human Body Configurations:

Combining Segmentation and Recognition” CVPR 2004

• Thank you!