object category detection: parts-based modelsdhoiem.cs.illinois.edu/courses/vision_spring10... ·...
TRANSCRIPT
![Page 1: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/1.jpg)
Object Category Detection:Parts-based Models
Computer Vision
CS 543 / ECE 549
University of Illinois
Derek Hoiem
03/30/10
![Page 2: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/2.jpg)
Administrative stuff
• Returning homeworks
![Page 3: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/3.jpg)
Administrative stuff• Projects: next class, each group gives 2 minute
summary of projects– Goal– Progress so far– If you want to show an image or figure, e-mail me one
image or powerpoint slide by Wed 5pm
• Deadlines– HW 3 due April 6– HW 4 due May 4 (should be out April 9)– Projects due in ~5 weeks
• Poster session– During finals week
• May 11 12:30-2:30 (Tues) or May 13 12:30-2:30 (Thurs)
![Page 4: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/4.jpg)
Goal: Detect all instances of objectsCars
Faces
Cats
![Page 5: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/5.jpg)
Object model: last class
• Statistical Template in Bounding Box– Object is some (x,y,w,h) in image
– Features defined wrt bounding box coordinates
Image Template Visualization
Images from Felzenszwalb
![Page 6: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/6.jpg)
Last class: sliding window detection
![Page 7: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/7.jpg)
Last class: statistical template
• Object model = log linear model of parts at fixed positions
+3 +2 -2 -1 -2.5 = -0.5
+4 +1 +0.5 +3 +0.5= 10.5
> 7.5?
> 7.5?
Non-object
Object
![Page 8: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/8.jpg)
When do statistical templates make sense?
Caltech 101 Average Object Images
![Page 9: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/9.jpg)
Object models: this class
• Articulated parts model– Object is configuration of parts
– Each part is detectable
Images from Felzenszwalb
![Page 10: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/10.jpg)
Deformable objects
Images from Caltech-256
Slide Credit: Duan Tran
![Page 11: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/11.jpg)
Deformable objects
Images from D. Ramanan’s datasetSlide Credit: Duan Tran
![Page 12: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/12.jpg)
Compositional objects
![Page 13: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/13.jpg)
Parts-based Models
Define object by collection of parts modeled by1. Appearance
2. Spatial configuration
Slide credit: Rob Fergus
![Page 14: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/14.jpg)
How to model spatial relations?
• One extreme: fixed template
![Page 15: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/15.jpg)
How to model spatial relations?
• Another extreme: bag of words
=
![Page 16: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/16.jpg)
How to model spatial relations?
• Star-shaped model
Root
Part
Part
Part
Part
Part
![Page 17: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/17.jpg)
How to model spatial relations?
• Star-shaped model
=X X
XRoot
Part
Part
Part
Part
Part
![Page 18: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/18.jpg)
How to model spatial relations?• Tree-shaped model
![Page 19: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/19.jpg)
How to model spatial relations?
Fergus et al. ’03Fei-Fei et al. ‘03
Leibe et al. ’04, ‘08Crandall et al. ‘05Fergus et al. ’05
Crandall et al. ‘05 Felzenszwalb & Huttenlocher ‘05
Bouchard & Triggs ‘05 Carneiro & Lowe ‘06Csurka ’04Vasconcelos ‘00
from [Carneiro & Lowe, ECCV’06]
O(N6) O(N2) O(N3) O(N2)
• Many others...
![Page 20: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/20.jpg)
Today’s class
1. Star-shaped model – Example: ISM
• Leibe et al. 2004, 2008
2. Tree-shaped model– Example: Pictorial structures
• Felzenszwalb Huttenlocher 2005
Root
Part
Part
Part
Part
Part
![Page 21: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/21.jpg)
ISM: Implicit Shape Model
Training overview• Start with bounding boxes and (ideally) segmentations of
objects
• Extract local features (e.g., patches or SIFT) at interest points on objects
• Cluster features to create codebook
• Record relative bounding box and segmentation for each codeword
![Page 22: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/22.jpg)
ISM: Implicit Shape Model
Testing overview• Extract interest points in test image
• Softly match to codebook entries
• Each matched codeword votes for object bounding box
• Compute modes of votes using mean-shift
• Check which codewords voted for modes
• Refine
![Page 23: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/23.jpg)
K. Grauman, B. Leibe
Codebook Representation
• Extraction of local object featuresInterest Points (e.g. Harris detector)Sparse representation of the object appearance
• Collect features from whole training set
• Example:
![Page 24: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/24.jpg)
K. Grauman, B. Leibe
Agglomerative Clustering
• Algorithm (Average-Link)1. Start with each patch as a cluster of its own2. Repeatedly merge the two most similar clusters X and Y,
where the similarity between two clusters is defined as the average similarity between their members
3. Until
• Commonly used similarity measuresNormalized correlationEuclidean distances
θ<),sim( YX
![Page 25: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/25.jpg)
K. Grauman, B. Leibe
Appearance Codebook
• Clustering ResultsVisual similarity preservedWheel parts, window corners, fenders, ...Store cluster centers as Appearance Codebook
…
![Page 26: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/26.jpg)
K. Grauman, B. Leibe
Voting with Local Features
• For every feature, store possible “occurrences”
• For new image, let the matched features vote for possible object positions
Record relative size and scale of object
![Page 27: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/27.jpg)
Implicit Shape Model - RecognitionInterest Points Matched Codebook
EntriesProbabilistic
Voting
3D Voting Space(continuous)
x
y
s
Object Position
o,x
Image Feature
f
Interpretation(Codebook match)
Ci
)( fCp i ),,( lin Cxop
∑=i
inin CxopfCpfxop ),,()(),,( ll
[Leibe04, Leibe08]
![Page 28: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/28.jpg)
K. Grauman, B. Leibe
• Mean-Shift formulation for refinementScale-adaptive balloon density estimator
Scale Voting: Efficient Computation
y
s
Binned accum. array
y
s
x
Refinement(MSME)
y
s
x
Candidatemaxima
y
s
Scale votes
![Page 29: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/29.jpg)
Implicit Shape Model - Recognition
BackprojectedHypotheses
Interest Points Matched Codebook Entries
Probabilistic Voting
3D Voting Space(continuous)
x
y
s
Backprojectionof Maxima
[Leibe04, Leibe08]
![Page 30: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/30.jpg)
K. Grauman, B. Leibe
Original image
Example: Results on Cows
![Page 31: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/31.jpg)
K. Grauman, B. Leibe
Original imageInterest points
Example: Results on Cows
![Page 32: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/32.jpg)
K. Grauman, B. Leibe
O������� �����I�������
������
Matched patches
Example: Results on Cows
![Page 33: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/33.jpg)
K. Grauman, B. Leibe
O������� �����I�������
������
M������ �������Prob. Votes
Example: Results on Cows
![Page 34: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/34.jpg)
K. Grauman, B. Leibe
1st hypothesis
Example: Results on Cows
![Page 35: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/35.jpg)
K. Grauman, B. Leibe
2nd hypothesis
Example: Results on Cows
![Page 36: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/36.jpg)
K. Grauman, B. Leibe
Example: Results on Cows
3rd hypothesis
![Page 37: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/37.jpg)
ISM: Detection Results
• Qualitative Performance– Recognizes different kinds of objects
– Robust to clutter, occlusion, noise, low contrast
K. Grauman, B. Leibe
![Page 38: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/38.jpg)
Beyond bounding boxes
Backprojected codewords can vote:• Pixel segmentation
• Part layout
• Pose
• Depth values BackprojectedHypotheses
Interest Points Matched Codebook Entries
Probabilistic Voting
3D Voting Space(continuous)
x
y
s
Backprojectionof Maxima
![Page 39: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/39.jpg)
K. Grauman, B. Leibe
Segmentation: Probabilistic Formulation
• Influence of patch on object hypothesis (vote weight)
( ) ( ) ( ) ( )( )xop
f,pfCpCxopxofp
n
i iinn ,
||,,, ∑= l
l
( ) ( ) ( )∑∈
===),(
,|,,,,|,|l
llf
nnn xofpxoffigurepxofigurepp
pp• Backprojection to features f and pixels p:
Segmentationinformation
Influence on object hypothesis
[Leibe04, Leibe08]
![Page 40: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/40.jpg)
K. Grauman, B. Leibe
ISM – Top-Down Segmentation
BackprojectedHypotheses
Interest Points Matched Codebook Entries
Probabilistic Voting
Segmentation3D Voting Space
(continuous)
x
y
s
Backprojectionof Maxima
p(figure)Probabilities
[Leibe04, Leibe08]
![Page 41: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/41.jpg)
46K. Grauman, B. Leibe
Example Results: Motorbikes
![Page 42: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/42.jpg)
47B. Leibe
Example Results: Chairs
Office chairs
Dining room chairs
![Page 43: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/43.jpg)
48
Inferring Other Information: Part Labels
Training
Test Output
[Thomas07]
![Page 44: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/44.jpg)
49
Inferring Other Information: Part Labels (2)
[Thomas07]
![Page 45: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/45.jpg)
50
Inferring Other Information: Depth Maps
“Depth from a single image”
[Thomas07]
![Page 46: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/46.jpg)
Tree-shaped model
![Page 47: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/47.jpg)
Pictorial Structures Model
Part = oriented rectangle Spatial model = relative size/orientation
Felzenszwalb and Huttenlocher 2005
![Page 48: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/48.jpg)
Pictorial Structures Model
Appearance likelihood Geometry likelihood
![Page 49: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/49.jpg)
Pictorial structures model
Optimization is tricky but can be efficient
Maximization
• For each l1, find best l2:
• Remove v2, and repeat with smaller tree, until only a single part
• For n parts, k locations per part, this has complexity of O(nk2), but can be solved in ~O(nk) usinggeneralized distance transform
![Page 50: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/50.jpg)
Pictorial structures model
Optimization is tricky but can be efficient
Sampling
• Sample root node, then each node given parent, until all parts are sampled
![Page 51: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/51.jpg)
Sample poses from likelihood and choose best match with Chamfer distance
![Page 52: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/52.jpg)
Results for person matching
58
![Page 53: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/53.jpg)
Results for person matching
59
![Page 54: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/54.jpg)
Recently enhanced pictorial structures
BMVC 2009
![Page 55: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/55.jpg)
Things to remember• Rather than searching for whole
object, can locate “parts” that vote for object– Better encoding of spatial
variation
• These parts can vote for other things too
• Models can be broken down into part appearance and spatial configuration– Wide variety of models
• Efficient optimization is often tricky, but many tricks available
![Page 56: Object Category Detection: Parts-based Modelsdhoiem.cs.illinois.edu/courses/vision_spring10... · Object Category Detection: Parts-based Models Computer Vision CS 543 / ECE 549](https://reader034.vdocuments.us/reader034/viewer/2022042922/5f6f33bfc3cf2143e6650540/html5/thumbnails/56.jpg)
Next class
• Each group gives 2 minute summary of projects– Goal
– Progress so far
– If you want to show an image or figure, e-mail me one image or powerpoint slide by Wed 5pm
• Review of object recognition