efficient part based recognition of multiple object classes n.
Download
Skip this Video
Loading SlideShow in 5 Seconds..
Efficient Part-Based Recognition of Multiple Object Classes PowerPoint Presentation
Download Presentation
Efficient Part-Based Recognition of Multiple Object Classes

Loading in 2 Seconds...

play fullscreen
1 / 172

Efficient Part-Based Recognition of Multiple Object Classes - PowerPoint PPT Presentation


  • 129 Views
  • Uploaded on

Efficient Part-Based Recognition of Multiple Object Classes. Object Class Recognition Systems. Model Representation. 2. Object Class Recognition Systems. Model Representation Learning Algorithm. 3. Object Class Recognition Systems. Model Representation Learning Algorithm

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about 'Efficient Part-Based Recognition of Multiple Object Classes' - curran-bradshaw


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
object class recognition systems1
Object Class Recognition Systems
  • Model Representation
  • Learning Algorithm

3

object class recognition systems2
Object Class Recognition Systems
  • Model Representation
  • Learning Algorithm
  • Recognition Algorithm

4

object class recognition systems3
Object Class Recognition Systems
  • Model Representation
    • Part-based model
  • Learning Algorithm
  • Recognition Algorithm

5

object class recognition systems4
Object Class Recognition Systems
  • Model Representation
    • Part-based model
      • Part appearance
  • Learning Algorithm
  • Recognition Algorithm

6

object class recognition systems5
Object Class Recognition Systems
  • Model Representation
    • Part-based model
      • Part appearance
      • Part position
  • Learning Algorithm
  • Recognition Algorithm

7

object class recognition systems6
Object Class Recognition Systems
  • Model Representation
    • Part-based model
      • Part appearance
      • Part position
  • Learning Algorithm
  • Recognition Algorithm

8

history of part based object recognition
History of Part-BasedObject Recognition

Hoffman 1999

Bag of Parts

Crandall et al. 2005, 2006

k-Fans

Proposed

Algorithm

Template

Matching

Part-Based

Burl et al. 1998

Constellation

Model

Fergus et al. 2003, 2005

9

template matching

BoP

k-Fan

this

Templ

Part

Const

Template Matching

10

template matching1

BoP

k-Fan

this

Templ

Part

Const

Template Matching

11

template matching2

BoP

k-Fan

this

Templ

Part

Const

Template Matching

12

template matching3

BoP

k-Fan

this

Templ

Part

Const

Template Matching

13

template matching4

BoP

k-Fan

this

Templ

Part

Const

Template Matching

14

template matching5

BoP

k-Fan

this

Templ

Part

Const

Template Matching

15

template matching6

BoP

k-Fan

this

Templ

Part

Const

Template Matching

16

template matching7

BoP

k-Fan

this

Templ

Part

Const

Template Matching

17

template matching8

BoP

k-Fan

this

Templ

Part

Const

Template Matching

18

template matching9

BoP

k-Fan

this

Templ

Part

Const

Template Matching

19

template matching10

BoP

k-Fan

this

Templ

Part

Const

Template Matching

20

template matching11

BoP

k-Fan

this

Templ

Part

Const

Template Matching

21

template matching12

BoP

k-Fan

this

Templ

Part

Const

Template Matching

22

template matching13

BoP

k-Fan

this

Templ

Part

Const

Template Matching

23

template matching14

BoP

k-Fan

this

Templ

Part

Const

Template Matching

24

template matching15

BoP

k-Fan

this

Templ

Part

Const

Template Matching

max

25

template matching16

BoP

k-Fan

this

Templ

Part

Const

Template Matching

26

template matching17

BoP

k-Fan

this

Templ

Part

Const

Template Matching

27

template matching18

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels)

28

template matching19

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels) – Yes, O(N)

29

template matching20

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels) – Yes, O(N)
  • Robust?

30

template matching21

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels) – Yes, O(N)
  • Robust?

31

template matching22

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels) – Yes, O(N)
  • Robust?

32

template matching23

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels) – Yes, O(N)
  • Robust?

33

template matching24

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels) – Yes, O(N)
  • Robust?

34

template matching25

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels) – Yes, O(N)
  • Robust?

35

template matching26

BoP

k-Fan

this

Templ

Part

Const

Template Matching
  • Efficient? (N = # pixels) – Yes, O(N)
  • Robust? – No, inflexible

36

template vs part based

BoP

k-Fan

this

Templ

Part

Const

Template vs. Part-Based

Robust

Efficient

Trade-off

37

part based

BoP

k-Fan

this

Templ

Part

Const

Part-Based

38

part based1

BoP

k-Fan

this

Templ

Part

Const

Part-Based

39

part based2

BoP

k-Fan

this

Templ

Part

Const

Part-Based

40

part based3

BoP

k-Fan

this

Templ

Part

Const

Part-Based

41

part based4

BoP

k-Fan

this

Templ

Part

Const

Part-Based

42

part based5

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)

43

part based6

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)

44

part based7

BoP

k-Fan

this

Templ

Part

Const

Part-Based

45

part based8

BoP

k-Fan

this

Templ

Part

Const

Part-Based

46

part based9

BoP

k-Fan

this

Templ

Part

Const

Part-Based

47

part based10

BoP

k-Fan

this

Templ

Part

Const

Part-Based

48

part based11

BoP

k-Fan

this

Templ

Part

Const

Part-Based

49

part based12

BoP

k-Fan

this

Templ

Part

Const

Part-Based

50

part based13

BoP

k-Fan

this

Templ

Part

Const

Part-Based

51

part based14

BoP

k-Fan

this

Templ

Part

Const

Part-Based

52

part based15

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels)

53

part based16

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels)

P

54

part based17

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels)

N

P

55

part based18

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels)

N

P

56

part based19

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels)

N

N

P

57

part based20

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels) – No, O(PN^P)

N

N

P

58

part based21

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels) – No, O(PN^P)
  • Robust?

N

N

P

59

part based22

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels) – No, O(PN^P)
  • Robust? Yes!

N

N

P

60

part based23

BoP

k-Fan

this

Templ

Part

Const

Part-Based
  • Part appearance (template matching)
  • Part location (fully connected)
  • Efficient? (P = #parts, N = #pixels) – No, O(PN^P)
  • Robust? Yes!
  • All following algorithms try to approach the accuracy of this method, while gaining efficiency

N

N

P

61

constellation model

BoP

k-Fan

this

Templ

Part

Const

Constellation Model

Dense Image (n=N)

Robust

Efficient

Trade-off

62

constellation model1

BoP

k-Fan

this

Templ

Part

Const

Constellation Model

Sparse Interest Points (n<<N)

Dense Image (n=N)

Robust

Efficient

Trade-off

63

constellation model2

BoP

k-Fan

this

Templ

Part

Const

Constellation Model
  • Efficient O(Pn^P) (n = # interest points, P = # parts)

64

constellation model3

BoP

k-Fan

this

Templ

Part

Const

Constellation Model
  • Efficient O(Pn^P) (n = # interest points, P = # parts)
  • Approximation:
    • Sparse image: only consider interest points (n<<N)

65

constellation model4

BoP

k-Fan

this

Templ

Part

Const

Constellation Model
  • Efficient O(Pn^P) (n = # interest points, P = # parts)
  • Approximation:
    • Sparse image: only consider interest points (n<<N)
  • Interest point detector too general
    • regions of the image discarded without considering particular parts that may be there

66

bag of parts

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

Robust

Efficient

Trade-off

67

bag of parts1

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

Robust

Efficient

Trade-off

68

bag of parts2

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

69

bag of parts3

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)

70

bag of parts4

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)

71

bag of parts5

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

72

bag of parts6

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

73

bag of parts7

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

74

bag of parts8

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

75

bag of parts9

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

76

bag of parts10

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts

77

bag of parts11

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)

max

max

78

bag of parts12

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels)

79

bag of parts13

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels)

N

80

bag of parts14

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels)

P

N

81

bag of parts15

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels) – Yes, O(NP)

P

N

82

bag of parts16

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels) – Yes, O(NP)
  • Robust?

83

bag of parts17

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels) – Yes, O(NP)
  • Robust?

84

bag of parts18

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels) – Yes, O(NP)
  • Robust?
    • No localization

85

bag of parts19

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels) – Yes, O(NP)
  • Robust?
    • No localization

86

bag of parts20

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels) – Yes, O(NP)
  • Robust?
    • No localization

87

bag of parts21

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels) – Yes, O(NP)
  • Robust?
    • No localization
    • More likely to find false detections

88

bag of parts22

BoP

k-Fan

this

Templ

Part

Const

Bag of Parts
  • Part appearance (template matching)
  • Part location (ignored, disconnected)
  • Efficient? (P = #parts, N = #pixels) – Yes, O(NP)
  • Robust?
    • No localization
    • More likely to find false detections
    • …but still in common use

89

k fans

BoP

k-Fan

this

Templ

Part

Const

k-Fans

Robust

Efficient

90

k fans1

BoP

k-Fan

this

Templ

Part

Const

k-Fans

Robust

Efficient

91

k fans2

BoP

k-Fan

this

Templ

Part

Const

k-Fans

Robust

Efficient

92

k fans3

BoP

k-Fan

this

Templ

Part

Const

k-Fans

Robust

Efficient

93

k fans4

BoP

k-Fan

this

Templ

Part

Const

k-Fans

Robust

Efficient

Trade-off

94

1 fans

BoP

k-Fan

this

Templ

Part

Const

1-Fans

95

1 fans1

BoP

k-Fan

this

Templ

Part

Const

1-Fans

96

1 fans2

BoP

k-Fan

this

Templ

Part

Const

1-Fans

97

1 fans3

BoP

k-Fan

this

Templ

Part

Const

1-Fans

98

1 fans4

BoP

k-Fan

this

Templ

Part

Const

1-Fans

99

1 fans5

BoP

k-Fan

this

Templ

Part

Const

1-Fans

100

1 fans6

BoP

k-Fan

this

Templ

Part

Const

1-Fans
  • Look at the probability model for locations

101

1 fans7

BoP

k-Fan

this

Templ

Part

Const

1-Fans
  • Look at the probability model for locations

102

1 fans8

BoP

k-Fan

this

Templ

Part

Const

1-Fans
  • Look at the probability model for locations

103

1 fans9

BoP

k-Fan

this

Templ

Part

Const

1-Fans

104

1 fans10

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

105

1 fans11

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

106

1 fans12

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

107

1 fans13

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

108

1 fans14

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

109

1 fans15

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

110

1 fans16

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

111

1 fans17

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

112

1 fans18

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

113

1 fans19

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

114

1 fans20

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

115

1 fans21

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

116

1 fans22

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

117

1 fans23

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

118

1 fans24

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

119

1 fans25

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

120

1 fans26

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

121

1 fans27

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

122

1 fans28

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

123

1 fans29

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

124

1 fans30

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

125

1 fans31

BoP

k-Fan

this

Templ

Part

Const

1-Fans

...

126

1 fans32

BoP

k-Fan

this

Templ

Part

Const

1-Fans
1 fan

BoP

k-Fan

this

Templ

Part

Const

1-Fan

+

128

1 fans33
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts)

129

1 fans34
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts)

N

130

1 fans35
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts)

P

N

131

1 fans36
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts)

N

P

N

132

1 fans37
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts) Yes, O(PN^2)

N

P

N

133

1 fans38
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts) Yes, O(PN^2)
    • Use Dynamic Programming:

N

P

N

134

1 fans39
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts) Yes, O(PN^2)
    • Use Dynamic Programming:
      • Precompute DTj image in O(N)

N

P

N

135

1 fans40
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts) Yes, O(PN^2)
    • Use Dynamic Programming:
      • Precompute DTj image in O(N)
      • Then add the A1 and DTj images

N

P

N

136

1 fans41
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts) Yes, O(PN^2) O(NP) – Same as bag of parts!
    • Use Dynamic Programming:
      • Precompute DTj image in O(N)
      • Then add the A1 and DTj images

N

P

N

137

1 fans42
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts) Yes, O(PN^2) O(NP) – Same as bag of parts!
  • Robust?

138

1 fans43
1-Fans

BoP

k-Fan

this

Templ

Part

Const

  • Efficient? (N = # pixels, P = # parts) Yes, O(PN^2) O(NP) – Same as bag of parts!
  • Robust? – Yes, better than bag of parts.

139

aren t we done
Aren’t We Done?
  • Object Recognition is efficient and robust. Can’t we stop here?
aren t we done1
Aren’t We Done?
  • Object Recognition is efficient and robust. Can’t we stop here?
  • What about…
aren t we done2
Aren’t We Done?
  • Object Recognition is efficient and robust. Can’t we stop here?
  • What about…
    • Detecting multiple object classes? (M = # objects… think 30,000) O(MNP)
recap history of part based object recognition
Recap: History of Part-BasedObject Recognition

Hoffman 1999

Bag of Parts

Crandall et al. 2005, 2006

k-Fans

Proposed

Algorithm

Template

Matching

Part-Based

Burl et al. 1998

Constellation

Model

Fergus et al. 2003, 2005

143

proposed algorithm

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

1-fan

Accurate

Efficient

Sparse Appearance Image

(thresholded)

Does not rely on ageneral

interest point detector

144

proposed algorithm1

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

145

proposed algorithm2

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

146

proposed algorithm3

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

This is sufficient

to solve 0-fans:

147

proposed algorithm4

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

max

max

This is sufficient

to solve 0-fans:

148

proposed algorithm5

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm
  • How do we threshold appearances in sublinear time (i.e. < O(MNP)? M = # objects, N = # pixels, P = # parts/object)
proposed algorithm6

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm
  • How do we threshold appearances in sublinear time (i.e. < O(MNP)? M = # objects, N = # pixels, P = # parts/object)
  • View thresholded appearance detection as an R-nearest neighbor problem
r nearest neighbors

BoP

k-Fan

this

Templ

Part

Const

R-Nearest Neighbors
  • What is the set of points in a database that are within a radius R from the query point q?

q

r nearest neighbors1

BoP

k-Fan

this

Templ

Part

Const

R-Nearest Neighbors
  • What is the set of points in a database that are within a radius R from the query point q?

R

q

r nearest neighbors2

BoP

k-Fan

this

Templ

Part

Const

R-Nearest Neighbors
  • What is the set of points in a database that are within a radius R from the query point q?

q

r nearest neighbors3
R-Nearest Neighbors
  • The space  part appearances (high dimensions)

q

r nearest neighbors4
R-Nearest Neighbors
  • The space  part appearances (high dimensions)
  • Assume part appearances are identically distributed spherical Gaussians (iΣi =cI)

q

r nearest neighbors5
R-Nearest Neighbors
  • The space  part appearances (high dimensions)
  • Assume part appearances are identically distributed spherical Gaussians (iΣi =cI)
  • Database points are the means (μi)

q

r nearest neighbors6
R-Nearest Neighbors
  • The space  part appearances (high dimensions)
  • Assume part appearances are identically distributed spherical Gaussians (iΣi =cI)
  • Database points are the means (μi)
  • 1-Fan part appearances can be expressed this way

q

r nearest neighbors7
R-Nearest Neighbors
  • Okay, how fast can it be solved?

q

r nearest neighbors8
R-Nearest Neighbors
  • Okay, how fast can it be solved?
    • In low dimensions (<8), kD trees solve it in sublinear time

q

r nearest neighbors9
R-Nearest Neighbors
  • Okay, how fast can it be solved?
    • In low dimensions (<8), kD trees solve it in sublinear time
    • But the conjectured “curse of dimensionality” prevents it from being solved efficiently for high dimensions

q

r nearest neighbors10
R-Nearest Neighbors
  • Okay, how fast can it be solved?
    • In low dimensions (<8), kD trees solve it in sublinear time
    • But the conjectured “curse of dimensionality” prevents it from being solved efficiently for high dimensions
  • Locality Sensitive Hashing solves the problem approximately,
    • misses some points with probability 1-δ
    • Solves it in O(nd1/c+o(1))
    • Trades off probability of false negative for efficiency
proposed algorithm7

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

162

proposed algorithm8

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

163

proposed algorithm9

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

164

proposed algorithm10

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

165

proposed algorithm11

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

166

proposed algorithm12

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

167

proposed algorithm13

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

168

proposed algorithm14

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

169

proposed algorithm15

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

Now we can do this efficiently

170

proposed algorithm16

BoP

k-Fan

this

Templ

Part

Const

Proposed Algorithm

...

How do we do these efficiently?

171

proposed algorithm17
Proposed Algorithm
  • Lazy distance transforms, and HA*LD
  • Coming soon 