SlideShare a Scribd company logo
1 of 33
Download to read offline
1/26
Deformable Part Models are Convolutional
Neural Networks
Ross Girshick, Forrest Iandola, Trevor Darrell, Jitendra Malik
Presentor: YANG Wei
January 25, 2016
2/26
Outline
1 Introduction
2 DeepPyramid DPMs
Feature pyramid front-end CNN
Constructing an equivalent CNN from a DPM
3 Implementation details
4 Experiments
3/26
Outline
1 Introduction
2 DeepPyramid DPMs
Feature pyramid front-end CNN
Constructing an equivalent CNN from a DPM
3 Implementation details
4 Experiments
4/26
Deformable Part Models vs. Convolutional Neural
Networks
Deformable part models
4/26
Deformable Part Models vs. Convolutional Neural
Networks
Deformable part models
Convolutional neural networks
5/26
Are DPMs and CNNs actually distinct?
DPMs: graphical models
CNNs: “black-box” non-linear classifiers
This paper shows that any DPM can be formulated as an
equivalent CNN, i.e., deformable part models are convolutional
neural networks.
6/26
Outline
1 Introduction
2 DeepPyramid DPMs
Feature pyramid front-end CNN
Constructing an equivalent CNN from a DPM
3 Implementation details
4 Experiments
7/26
DeepPyramid DPMs
Schematic model overview: “front-end CNN” + DPM-CNN
input: image pyramid
output: object detection scores
8/26
Feature pyramid front-end CNN
front-end CNN: AlexNet (conv1-conv5).
A CNN that maps an image pyramid to a feature pyramid
AlexNet
single-scale architecture
9/26
Constructing an equivalent CNN from a DPM
A single-component DPM.
mixture of components
component = root filter + part filter
10/26
Inference with DPMs
The matching process at one scale.
11/26
Architecture of DPM-CNN
The unrolled detection algorithm of DPM generates a specific
network with fixed length:
1 input: conv5 feature pyramid of front-end CNN
11/26
Architecture of DPM-CNN
The unrolled detection algorithm of DPM generates a specific
network with fixed length:
1 input: conv5 feature pyramid of front-end CNN
2 generate P+1 feature maps: 1 root filter and P part filters
11/26
Architecture of DPM-CNN
The unrolled detection algorithm of DPM generates a specific
network with fixed length:
1 input: conv5 feature pyramid of front-end CNN
2 generate P+1 feature maps: 1 root filter and P part filters
3 P part feature maps are fed into distance transform layer
11/26
Architecture of DPM-CNN
The unrolled detection algorithm of DPM generates a specific
network with fixed length:
1 input: conv5 feature pyramid of front-end CNN
2 generate P+1 feature maps: 1 root filter and P part filters
3 P part feature maps are fed into distance transform layer
4 root feature map are stacked (channel-wise concatenated)
with the transformed part feature maps
11/26
Architecture of DPM-CNN
The unrolled detection algorithm of DPM generates a specific
network with fixed length:
1 input: conv5 feature pyramid of front-end CNN
2 generate P+1 feature maps: 1 root filter and P part filters
3 P part feature maps are fed into distance transform layer
4 root feature map are stacked (channel-wise concatenated)
with the transformed part feature maps
5 The resulting P+1 channel feature map is convolved with
an object geometry filter, which produces the output DPM
score map for the input pyramid level
12/26
Architecture of DPM-CNN
CNN equivalent to a single-component DPM.
13/26
Traditional distance transform
Traditional distance transforms are defined for sets of points on
a grid [FH05].
G: grid
d(p−q): measure of
distance between points
p,q ∈ G
B ⊆ G
Then the distance transform of
B on G
DB(p) = min
q∈B
d(p−q)
Distance transform (Euclidean distance)
14/26
Traditional distance transform
DT can be also formulated as
DB(p) = min
q∈G
(d(p−q)+1B(q))
where
1B(q) =
0, if q ∈ B,
∞, if q /∈ B.
(1)
15/26
Generalized distance transform
A generalization of distance transforms can be obtained by
replacing the indicator function with some arbitrary function
over the grid G
Df (p) = min
q∈G
(d(p−q)+ f (q))
We can also define the generalized DT as maximization by
letting f(q) = −f (q)
Df (p) = max
q∈G
(f(q)−d(p−q))
16/26
Distance transform in DPM
In DPM, after computing filter responses we transform the
responses of the part filters to allow spatial uncertainty,
Di(x,y) = max
dx,dy
(Ri(x+dx,y+dy)−wi ·φd(dx,dy))
where
φd(dx,dy) = [dx,dy,dx2
,,dy2
]
The value Di(x,y) is the maximum contribution of the part
to the score of a root location that places the anchor of this
part at position (x,y).
16/26
Distance transform in DPM
In DPM, after computing filter responses we transform the
responses of the part filters to allow spatial uncertainty,
Di(x,y) = max
dx,dy
(Ri(x+dx,y+dy)−wi ·φd(dx,dy))
where
φd(dx,dy) = [dx,dy,dx2
,,dy2
]
The value Di(x,y) is the maximum contribution of the part
to the score of a root location that places the anchor of this
part at position (x,y).
By letting p = (x,y), p−q = (dx,dy) and
d(p−q) = w·φ(p−q), we can see that it is exactly in the
form of distance transform.
17/26
Max pooling as distance transform
Consider max pooling on f : G → R on a regular grid G.
Let a window half-length as k, then max pooling can be defined
as
Mf (p) = max
∆p∈{−k,···,k}
f(p+∆p)
Max pooling can be expressed equivalently as distance
transform:
Mf (p) = max
q∈G
(f(q)−dmax(p−q))
where
dmax(p−q) =
0, if (p−q) ∈ {−k,··· ,k},
∞, otherwise .
(2)
18/26
Generalize max pooling to distance transform pooling
We can generalize max pooling to distance transform pooling:
unlike max pooling, the distance transform of f at p is
taken over the entire domain G
rather than specifying a fixed pooling window a priori, the
shape of the pooling region can be learned from the data.
18/26
Generalize max pooling to distance transform pooling
We can generalize max pooling to distance transform pooling:
unlike max pooling, the distance transform of f at p is
taken over the entire domain G
rather than specifying a fixed pooling window a priori, the
shape of the pooling region can be learned from the data.
The released code does not include the DT pooling layer.
Please refer to [OW13] for more details.
19/26
Object geometry filters
The root convolution map and the DT pooled part convolution maps are stacked into a
20/26
Combining mixture components with maxout
CNN equivalent to a multi-component DPM. A multi-component DPM-CNN is
composed of one DPM-CNN per component and a maxout [GWFM+13] layer that
takes a max over component DPM-CNN outputs at each location.
21/26
Outline
1 Introduction
2 DeepPyramid DPMs
Feature pyramid front-end CNN
Constructing an equivalent CNN from a DPM
3 Implementation details
4 Experiments
22/26
Feature pyramid front-end CNN
Implementation details
pretrain on ILSVRC 2012 classification using Caffe
use conv5 as output layer
“same” convolution
zero-pad each conv/pooling layer’s input with k/2 zeros
on all sides (top, bottom, left and right)
(x,y) in conv5 feature map has a receptive field centered on
pixel (16x,16y) in the input image
conv5 feature maps: stride: 16; receptive field: 163×163
23/26
Outline
1 Introduction
2 DeepPyramid DPMs
Feature pyramid front-end CNN
Constructing an equivalent CNN from a DPM
3 Implementation details
4 Experiments
24/26
Experiments
Detection average precision (%) on VOC 2007 test. Column C shows the number of
components and column P shows the number of parts per component.
25/26
Experiments
HOG versus conv5 feature pyramids. In contrast to HOG features, conv5 features are
more part-like and scale selective. Each conv5 pyramid shows 1 of 256 feature
channels. The top two rows show a HOG feature pyramid and the face channel of a
conv5 pyramid on the same input image.
26/26
References
Pedro F Felzenszwalb and Daniel P Huttenlocher, Pictorial structures for object
recognition, International Journal of Computer Vision 61 (2005), no. 1, 55–79.
Ian J Goodfellow, David Warde-Farley, Mehdi Mirza, Aaron Courville, and Yoshua
Bengio, Maxout networks, arXiv preprint arXiv:1302.4389 (2013).
Wanli Ouyang and Xiaogang Wang, Joint deep learning for pedestrian detection,
ICCV, IEEE, 2013, pp. 2056–2063.

More Related Content

What's hot

VECTOR VS PIECEWISE-LINEAR FITTING FOR SIGNAL AND POWER INTEGRITY SIMULATION
VECTOR VS PIECEWISE-LINEAR FITTING FOR SIGNAL AND POWER INTEGRITY SIMULATIONVECTOR VS PIECEWISE-LINEAR FITTING FOR SIGNAL AND POWER INTEGRITY SIMULATION
VECTOR VS PIECEWISE-LINEAR FITTING FOR SIGNAL AND POWER INTEGRITY SIMULATIONPiero Belforte
 
Report AdvancedCodingFinal - Pietro Santoro
Report AdvancedCodingFinal - Pietro SantoroReport AdvancedCodingFinal - Pietro Santoro
Report AdvancedCodingFinal - Pietro SantoroPietro Santoro
 
Motion estimation overview
Motion estimation overviewMotion estimation overview
Motion estimation overviewYoss Cohen
 
09 23sept 8434 10235-1-ed performance (edit ari)update 17jan18tyas
09 23sept 8434 10235-1-ed performance (edit ari)update 17jan18tyas09 23sept 8434 10235-1-ed performance (edit ari)update 17jan18tyas
09 23sept 8434 10235-1-ed performance (edit ari)update 17jan18tyasIAESIJEECS
 
Mat lab for bplc
Mat lab for bplcMat lab for bplc
Mat lab for bplcwendye13
 
Multiuser MIMO Vector Perturbation Precoding
Multiuser MIMO Vector Perturbation PrecodingMultiuser MIMO Vector Perturbation Precoding
Multiuser MIMO Vector Perturbation Precodingadeelrazi
 
A comparative study of different multiplier designs
A comparative study of different multiplier designsA comparative study of different multiplier designs
A comparative study of different multiplier designsHoopeer Hoopeer
 
Mobile radio chaneel matlab kostov
Mobile radio chaneel matlab kostovMobile radio chaneel matlab kostov
Mobile radio chaneel matlab kostovDwi Putra Asana
 
06 Bandwidth Utilization_Multiplexing_and_Spreading
06 Bandwidth Utilization_Multiplexing_and_Spreading06 Bandwidth Utilization_Multiplexing_and_Spreading
06 Bandwidth Utilization_Multiplexing_and_SpreadingAhmar Hashmi
 
Design and Fabrication of a Two Axis Parabolic Solar Dish Collector
Design and Fabrication of a Two Axis Parabolic Solar Dish CollectorDesign and Fabrication of a Two Axis Parabolic Solar Dish Collector
Design and Fabrication of a Two Axis Parabolic Solar Dish CollectorIJERA Editor
 
Low Complexity Multi-User MIMO Detection for Uplink SCMA System Using Expecta...
Low Complexity Multi-User MIMO Detection for Uplink SCMA System Using Expecta...Low Complexity Multi-User MIMO Detection for Uplink SCMA System Using Expecta...
Low Complexity Multi-User MIMO Detection for Uplink SCMA System Using Expecta...TELKOMNIKA JOURNAL
 
ALGORITHMS FOR PACKET ROUTING IN SWITCHING NETWORKS WITH RECONFIGURATION OVER...
ALGORITHMS FOR PACKET ROUTING IN SWITCHING NETWORKS WITH RECONFIGURATION OVER...ALGORITHMS FOR PACKET ROUTING IN SWITCHING NETWORKS WITH RECONFIGURATION OVER...
ALGORITHMS FOR PACKET ROUTING IN SWITCHING NETWORKS WITH RECONFIGURATION OVER...csandit
 
Physical layer network coding
Physical layer network codingPhysical layer network coding
Physical layer network codingNguyen Tan
 

What's hot (20)

VECTOR VS PIECEWISE-LINEAR FITTING FOR SIGNAL AND POWER INTEGRITY SIMULATION
VECTOR VS PIECEWISE-LINEAR FITTING FOR SIGNAL AND POWER INTEGRITY SIMULATIONVECTOR VS PIECEWISE-LINEAR FITTING FOR SIGNAL AND POWER INTEGRITY SIMULATION
VECTOR VS PIECEWISE-LINEAR FITTING FOR SIGNAL AND POWER INTEGRITY SIMULATION
 
Report AdvancedCodingFinal - Pietro Santoro
Report AdvancedCodingFinal - Pietro SantoroReport AdvancedCodingFinal - Pietro Santoro
Report AdvancedCodingFinal - Pietro Santoro
 
Ch 13
Ch 13Ch 13
Ch 13
 
Lecture set 5
Lecture set 5Lecture set 5
Lecture set 5
 
Motion estimation overview
Motion estimation overviewMotion estimation overview
Motion estimation overview
 
09 23sept 8434 10235-1-ed performance (edit ari)update 17jan18tyas
09 23sept 8434 10235-1-ed performance (edit ari)update 17jan18tyas09 23sept 8434 10235-1-ed performance (edit ari)update 17jan18tyas
09 23sept 8434 10235-1-ed performance (edit ari)update 17jan18tyas
 
Mat lab for bplc
Mat lab for bplcMat lab for bplc
Mat lab for bplc
 
Multiuser MIMO Vector Perturbation Precoding
Multiuser MIMO Vector Perturbation PrecodingMultiuser MIMO Vector Perturbation Precoding
Multiuser MIMO Vector Perturbation Precoding
 
A comparative study of different multiplier designs
A comparative study of different multiplier designsA comparative study of different multiplier designs
A comparative study of different multiplier designs
 
Mobile radio chaneel matlab kostov
Mobile radio chaneel matlab kostovMobile radio chaneel matlab kostov
Mobile radio chaneel matlab kostov
 
Dcp project
Dcp projectDcp project
Dcp project
 
Modulation techniques matlab_code
Modulation techniques matlab_codeModulation techniques matlab_code
Modulation techniques matlab_code
 
9517cnc05
9517cnc059517cnc05
9517cnc05
 
06 Bandwidth Utilization_Multiplexing_and_Spreading
06 Bandwidth Utilization_Multiplexing_and_Spreading06 Bandwidth Utilization_Multiplexing_and_Spreading
06 Bandwidth Utilization_Multiplexing_and_Spreading
 
Design and Fabrication of a Two Axis Parabolic Solar Dish Collector
Design and Fabrication of a Two Axis Parabolic Solar Dish CollectorDesign and Fabrication of a Two Axis Parabolic Solar Dish Collector
Design and Fabrication of a Two Axis Parabolic Solar Dish Collector
 
Low Complexity Multi-User MIMO Detection for Uplink SCMA System Using Expecta...
Low Complexity Multi-User MIMO Detection for Uplink SCMA System Using Expecta...Low Complexity Multi-User MIMO Detection for Uplink SCMA System Using Expecta...
Low Complexity Multi-User MIMO Detection for Uplink SCMA System Using Expecta...
 
Masters Report 3
Masters Report 3Masters Report 3
Masters Report 3
 
ALGORITHMS FOR PACKET ROUTING IN SWITCHING NETWORKS WITH RECONFIGURATION OVER...
ALGORITHMS FOR PACKET ROUTING IN SWITCHING NETWORKS WITH RECONFIGURATION OVER...ALGORITHMS FOR PACKET ROUTING IN SWITCHING NETWORKS WITH RECONFIGURATION OVER...
ALGORITHMS FOR PACKET ROUTING IN SWITCHING NETWORKS WITH RECONFIGURATION OVER...
 
Mimo
MimoMimo
Mimo
 
Physical layer network coding
Physical layer network codingPhysical layer network coding
Physical layer network coding
 

Viewers also liked

Deep learning-for-pose-estimation-wyang-defense
Deep learning-for-pose-estimation-wyang-defenseDeep learning-for-pose-estimation-wyang-defense
Deep learning-for-pose-estimation-wyang-defenseWei Yang
 
Pose Machine
Pose MachinePose Machine
Pose MachineWei Yang
 
Articulated human pose estimation by deep learning
Articulated human pose estimation by deep learningArticulated human pose estimation by deep learning
Articulated human pose estimation by deep learningWei Yang
 
Deep convolutional neural fields for depth estimation from a single image
Deep convolutional neural fields for depth estimation from a single imageDeep convolutional neural fields for depth estimation from a single image
Deep convolutional neural fields for depth estimation from a single imageWei Yang
 
Human Pose Estimation by Deep Learning
Human Pose Estimation by Deep LearningHuman Pose Estimation by Deep Learning
Human Pose Estimation by Deep LearningWei Yang
 
Manifold learning
Manifold learningManifold learning
Manifold learningWei Yang
 
Object Detection with Discrmininatively Trained Part based Models
Object Detection with Discrmininatively Trained Part based ModelsObject Detection with Discrmininatively Trained Part based Models
Object Detection with Discrmininatively Trained Part based Modelszukun
 
Real time pedestrian detection with deformable part models [h. cho, p. rybski...
Real time pedestrian detection with deformable part models [h. cho, p. rybski...Real time pedestrian detection with deformable part models [h. cho, p. rybski...
Real time pedestrian detection with deformable part models [h. cho, p. rybski...tino
 
High Performance Pedestrian Detection On TEGRA X1
High Performance Pedestrian Detection On TEGRA X1High Performance Pedestrian Detection On TEGRA X1
High Performance Pedestrian Detection On TEGRA X1NVIDIA
 
Andrey V. Savchenko - Sequential Hierarchical Image Recognition based on the ...
Andrey V. Savchenko - Sequential Hierarchical Image Recognition based on the ...Andrey V. Savchenko - Sequential Hierarchical Image Recognition based on the ...
Andrey V. Savchenko - Sequential Hierarchical Image Recognition based on the ...AIST
 
20 Instagram Pics that will Have You Wanting to Visit the Grand Canyon Just ...
20 Instagram Pics that will Have You Wanting to Visit the  Grand Canyon Just ...20 Instagram Pics that will Have You Wanting to Visit the  Grand Canyon Just ...
20 Instagram Pics that will Have You Wanting to Visit the Grand Canyon Just ...Grand Canyon Visitor Center
 
Docking Pose Assessment: The importance of keeping your GARD up
Docking Pose Assessment: The importance of keeping your GARD upDocking Pose Assessment: The importance of keeping your GARD up
Docking Pose Assessment: The importance of keeping your GARD upDavid Thompson
 
Learning Object Detectors From Weakly Supervised Image Data
Learning Object Detectors From Weakly Supervised Image DataLearning Object Detectors From Weakly Supervised Image Data
Learning Object Detectors From Weakly Supervised Image DataYandex
 
All pose face alignment robust to occlusion
All pose face alignment robust to occlusionAll pose face alignment robust to occlusion
All pose face alignment robust to occlusionJongju Shin
 
[Mmlab seminar 2016] deep learning for human pose estimation
[Mmlab seminar 2016] deep learning for human pose estimation[Mmlab seminar 2016] deep learning for human pose estimation
[Mmlab seminar 2016] deep learning for human pose estimationWei Yang
 
How to Get My Paper Accepted at Top Software Engineering Conferences
How to Get My Paper Accepted at Top Software Engineering ConferencesHow to Get My Paper Accepted at Top Software Engineering Conferences
How to Get My Paper Accepted at Top Software Engineering ConferencesAlex Orso
 
Pedestrian Detection Technology - Brochure
Pedestrian Detection Technology - BrochurePedestrian Detection Technology - Brochure
Pedestrian Detection Technology - BrochureMobileye
 
Single person pose recognition and tracking
Single person pose recognition and trackingSingle person pose recognition and tracking
Single person pose recognition and trackingJavier_Barbadillo
 

Viewers also liked (20)

Deep learning-for-pose-estimation-wyang-defense
Deep learning-for-pose-estimation-wyang-defenseDeep learning-for-pose-estimation-wyang-defense
Deep learning-for-pose-estimation-wyang-defense
 
Pose Machine
Pose MachinePose Machine
Pose Machine
 
Articulated human pose estimation by deep learning
Articulated human pose estimation by deep learningArticulated human pose estimation by deep learning
Articulated human pose estimation by deep learning
 
Deep convolutional neural fields for depth estimation from a single image
Deep convolutional neural fields for depth estimation from a single imageDeep convolutional neural fields for depth estimation from a single image
Deep convolutional neural fields for depth estimation from a single image
 
Human Pose Estimation by Deep Learning
Human Pose Estimation by Deep LearningHuman Pose Estimation by Deep Learning
Human Pose Estimation by Deep Learning
 
Manifold learning
Manifold learningManifold learning
Manifold learning
 
Object Detection with Discrmininatively Trained Part based Models
Object Detection with Discrmininatively Trained Part based ModelsObject Detection with Discrmininatively Trained Part based Models
Object Detection with Discrmininatively Trained Part based Models
 
Real time pedestrian detection with deformable part models [h. cho, p. rybski...
Real time pedestrian detection with deformable part models [h. cho, p. rybski...Real time pedestrian detection with deformable part models [h. cho, p. rybski...
Real time pedestrian detection with deformable part models [h. cho, p. rybski...
 
High Performance Pedestrian Detection On TEGRA X1
High Performance Pedestrian Detection On TEGRA X1High Performance Pedestrian Detection On TEGRA X1
High Performance Pedestrian Detection On TEGRA X1
 
Andrey V. Savchenko - Sequential Hierarchical Image Recognition based on the ...
Andrey V. Savchenko - Sequential Hierarchical Image Recognition based on the ...Andrey V. Savchenko - Sequential Hierarchical Image Recognition based on the ...
Andrey V. Savchenko - Sequential Hierarchical Image Recognition based on the ...
 
Monocular Human Pose Estimation with Bayesian Networks
Monocular Human Pose Estimation with Bayesian NetworksMonocular Human Pose Estimation with Bayesian Networks
Monocular Human Pose Estimation with Bayesian Networks
 
20 Instagram Pics that will Have You Wanting to Visit the Grand Canyon Just ...
20 Instagram Pics that will Have You Wanting to Visit the  Grand Canyon Just ...20 Instagram Pics that will Have You Wanting to Visit the  Grand Canyon Just ...
20 Instagram Pics that will Have You Wanting to Visit the Grand Canyon Just ...
 
Docking Pose Assessment: The importance of keeping your GARD up
Docking Pose Assessment: The importance of keeping your GARD upDocking Pose Assessment: The importance of keeping your GARD up
Docking Pose Assessment: The importance of keeping your GARD up
 
Learning Object Detectors From Weakly Supervised Image Data
Learning Object Detectors From Weakly Supervised Image DataLearning Object Detectors From Weakly Supervised Image Data
Learning Object Detectors From Weakly Supervised Image Data
 
All pose face alignment robust to occlusion
All pose face alignment robust to occlusionAll pose face alignment robust to occlusion
All pose face alignment robust to occlusion
 
Towards the Extended Pose
Towards the Extended PoseTowards the Extended Pose
Towards the Extended Pose
 
[Mmlab seminar 2016] deep learning for human pose estimation
[Mmlab seminar 2016] deep learning for human pose estimation[Mmlab seminar 2016] deep learning for human pose estimation
[Mmlab seminar 2016] deep learning for human pose estimation
 
How to Get My Paper Accepted at Top Software Engineering Conferences
How to Get My Paper Accepted at Top Software Engineering ConferencesHow to Get My Paper Accepted at Top Software Engineering Conferences
How to Get My Paper Accepted at Top Software Engineering Conferences
 
Pedestrian Detection Technology - Brochure
Pedestrian Detection Technology - BrochurePedestrian Detection Technology - Brochure
Pedestrian Detection Technology - Brochure
 
Single person pose recognition and tracking
Single person pose recognition and trackingSingle person pose recognition and tracking
Single person pose recognition and tracking
 

Similar to Deformable Part Models are Convolutional Neural Networks

Chapter 5 - Computer Networking a top-down Approach 7th
Chapter 5 - Computer Networking a top-down Approach 7thChapter 5 - Computer Networking a top-down Approach 7th
Chapter 5 - Computer Networking a top-down Approach 7thAndy Juan Sarango Veliz
 
Module 3- transport_layer .pptx
Module 3- transport_layer           .pptxModule 3- transport_layer           .pptx
Module 3- transport_layer .pptxhariprasad279825
 
International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)IJERD Editor
 
Computer graphics mini project on bellman-ford algorithm
Computer graphics mini project on bellman-ford algorithmComputer graphics mini project on bellman-ford algorithm
Computer graphics mini project on bellman-ford algorithmRAJEEV KUMAR SINGH
 
Parallel implementation of geodesic distance transform with application in su...
Parallel implementation of geodesic distance transform with application in su...Parallel implementation of geodesic distance transform with application in su...
Parallel implementation of geodesic distance transform with application in su...Tuan Q. Pham
 
Design Ofdm System And Remove Nonlinear Distortion In OFDM Signal At Transmit...
Design Ofdm System And Remove Nonlinear Distortion In OFDM Signal At Transmit...Design Ofdm System And Remove Nonlinear Distortion In OFDM Signal At Transmit...
Design Ofdm System And Remove Nonlinear Distortion In OFDM Signal At Transmit...Rupesh Sharma
 
Single Image Depth Estimation using frequency domain analysis and Deep learning
Single Image Depth Estimation using frequency domain analysis and Deep learningSingle Image Depth Estimation using frequency domain analysis and Deep learning
Single Image Depth Estimation using frequency domain analysis and Deep learningAhan M R
 
Employing non-orthogonal multiple access scheme in UAV-based wireless networks
Employing non-orthogonal multiple access scheme in UAV-based wireless networksEmploying non-orthogonal multiple access scheme in UAV-based wireless networks
Employing non-orthogonal multiple access scheme in UAV-based wireless networksjournalBEEI
 
Computer Graphics Notes 2.pdf
Computer Graphics Notes 2.pdfComputer Graphics Notes 2.pdf
Computer Graphics Notes 2.pdfAOUNHAIDER7
 
Attention is all you need (UPC Reading Group 2018, by Santi Pascual)
Attention is all you need (UPC Reading Group 2018, by Santi Pascual)Attention is all you need (UPC Reading Group 2018, by Santi Pascual)
Attention is all you need (UPC Reading Group 2018, by Santi Pascual)Universitat Politècnica de Catalunya
 
Analysis of Peak to Average Power Ratio Reduction Techniques in Sfbc Ofdm System
Analysis of Peak to Average Power Ratio Reduction Techniques in Sfbc Ofdm SystemAnalysis of Peak to Average Power Ratio Reduction Techniques in Sfbc Ofdm System
Analysis of Peak to Average Power Ratio Reduction Techniques in Sfbc Ofdm SystemIOSR Journals
 
IMPROVEMENT OF LTE DOWNLINK SYSTEM PERFORMANCES USING THE LAGRANGE POLYNOMIAL...
IMPROVEMENT OF LTE DOWNLINK SYSTEM PERFORMANCES USING THE LAGRANGE POLYNOMIAL...IMPROVEMENT OF LTE DOWNLINK SYSTEM PERFORMANCES USING THE LAGRANGE POLYNOMIAL...
IMPROVEMENT OF LTE DOWNLINK SYSTEM PERFORMANCES USING THE LAGRANGE POLYNOMIAL...IJCNCJournal
 

Similar to Deformable Part Models are Convolutional Neural Networks (20)

Chapter 5 - Computer Networking a top-down Approach 7th
Chapter 5 - Computer Networking a top-down Approach 7thChapter 5 - Computer Networking a top-down Approach 7th
Chapter 5 - Computer Networking a top-down Approach 7th
 
Module 3- transport_layer .pptx
Module 3- transport_layer           .pptxModule 3- transport_layer           .pptx
Module 3- transport_layer .pptx
 
International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)
 
Computer graphics mini project on bellman-ford algorithm
Computer graphics mini project on bellman-ford algorithmComputer graphics mini project on bellman-ford algorithm
Computer graphics mini project on bellman-ford algorithm
 
Parallel implementation of geodesic distance transform with application in su...
Parallel implementation of geodesic distance transform with application in su...Parallel implementation of geodesic distance transform with application in su...
Parallel implementation of geodesic distance transform with application in su...
 
ECE 565 FInal Project
ECE 565 FInal ProjectECE 565 FInal Project
ECE 565 FInal Project
 
Design Ofdm System And Remove Nonlinear Distortion In OFDM Signal At Transmit...
Design Ofdm System And Remove Nonlinear Distortion In OFDM Signal At Transmit...Design Ofdm System And Remove Nonlinear Distortion In OFDM Signal At Transmit...
Design Ofdm System And Remove Nonlinear Distortion In OFDM Signal At Transmit...
 
Single Image Depth Estimation using frequency domain analysis and Deep learning
Single Image Depth Estimation using frequency domain analysis and Deep learningSingle Image Depth Estimation using frequency domain analysis and Deep learning
Single Image Depth Estimation using frequency domain analysis and Deep learning
 
CNN.pptx
CNN.pptxCNN.pptx
CNN.pptx
 
Chapter 4 pc
Chapter 4 pcChapter 4 pc
Chapter 4 pc
 
www.ijerd.com
www.ijerd.comwww.ijerd.com
www.ijerd.com
 
Employing non-orthogonal multiple access scheme in UAV-based wireless networks
Employing non-orthogonal multiple access scheme in UAV-based wireless networksEmploying non-orthogonal multiple access scheme in UAV-based wireless networks
Employing non-orthogonal multiple access scheme in UAV-based wireless networks
 
Computer Graphics Notes 2.pdf
Computer Graphics Notes 2.pdfComputer Graphics Notes 2.pdf
Computer Graphics Notes 2.pdf
 
Attention is all you need (UPC Reading Group 2018, by Santi Pascual)
Attention is all you need (UPC Reading Group 2018, by Santi Pascual)Attention is all you need (UPC Reading Group 2018, by Santi Pascual)
Attention is all you need (UPC Reading Group 2018, by Santi Pascual)
 
4g lte matlab
4g lte matlab4g lte matlab
4g lte matlab
 
Skyline queries
Skyline queriesSkyline queries
Skyline queries
 
Analysis of Peak to Average Power Ratio Reduction Techniques in Sfbc Ofdm System
Analysis of Peak to Average Power Ratio Reduction Techniques in Sfbc Ofdm SystemAnalysis of Peak to Average Power Ratio Reduction Techniques in Sfbc Ofdm System
Analysis of Peak to Average Power Ratio Reduction Techniques in Sfbc Ofdm System
 
IMPROVEMENT OF LTE DOWNLINK SYSTEM PERFORMANCES USING THE LAGRANGE POLYNOMIAL...
IMPROVEMENT OF LTE DOWNLINK SYSTEM PERFORMANCES USING THE LAGRANGE POLYNOMIAL...IMPROVEMENT OF LTE DOWNLINK SYSTEM PERFORMANCES USING THE LAGRANGE POLYNOMIAL...
IMPROVEMENT OF LTE DOWNLINK SYSTEM PERFORMANCES USING THE LAGRANGE POLYNOMIAL...
 
ofdma doppler
ofdma dopplerofdma doppler
ofdma doppler
 
D04561722
D04561722D04561722
D04561722
 

Recently uploaded

GenBio2 - Lesson 1 - Introduction to Genetics.pptx
GenBio2 - Lesson 1 - Introduction to Genetics.pptxGenBio2 - Lesson 1 - Introduction to Genetics.pptx
GenBio2 - Lesson 1 - Introduction to Genetics.pptxBerniceCayabyab1
 
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptxRESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptxFarihaAbdulRasheed
 
Forest laws, Indian forest laws, why they are important
Forest laws, Indian forest laws, why they are importantForest laws, Indian forest laws, why they are important
Forest laws, Indian forest laws, why they are importantadityabhardwaj282
 
Artificial Intelligence In Microbiology by Dr. Prince C P
Artificial Intelligence In Microbiology by Dr. Prince C PArtificial Intelligence In Microbiology by Dr. Prince C P
Artificial Intelligence In Microbiology by Dr. Prince C PPRINCE C P
 
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxSOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxkessiyaTpeter
 
Evidences of Evolution General Biology 2
Evidences of Evolution General Biology 2Evidences of Evolution General Biology 2
Evidences of Evolution General Biology 2John Carlo Rollon
 
TOPIC 8 Temperature and Heat.pdf physics
TOPIC 8 Temperature and Heat.pdf physicsTOPIC 8 Temperature and Heat.pdf physics
TOPIC 8 Temperature and Heat.pdf physicsssuserddc89b
 
Transposable elements in prokaryotes.ppt
Transposable elements in prokaryotes.pptTransposable elements in prokaryotes.ppt
Transposable elements in prokaryotes.pptArshadWarsi13
 
OECD bibliometric indicators: Selected highlights, April 2024
OECD bibliometric indicators: Selected highlights, April 2024OECD bibliometric indicators: Selected highlights, April 2024
OECD bibliometric indicators: Selected highlights, April 2024innovationoecd
 
Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.
Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.
Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.aasikanpl
 
Pests of soyabean_Binomics_IdentificationDr.UPR.pdf
Pests of soyabean_Binomics_IdentificationDr.UPR.pdfPests of soyabean_Binomics_IdentificationDr.UPR.pdf
Pests of soyabean_Binomics_IdentificationDr.UPR.pdfPirithiRaju
 
Solution chemistry, Moral and Normal solutions
Solution chemistry, Moral and Normal solutionsSolution chemistry, Moral and Normal solutions
Solution chemistry, Moral and Normal solutionsHajira Mahmood
 
Twin's paradox experiment is a meassurement of the extra dimensions.pptx
Twin's paradox experiment is a meassurement of the extra dimensions.pptxTwin's paradox experiment is a meassurement of the extra dimensions.pptx
Twin's paradox experiment is a meassurement of the extra dimensions.pptxEran Akiva Sinbar
 
Harmful and Useful Microorganisms Presentation
Harmful and Useful Microorganisms PresentationHarmful and Useful Microorganisms Presentation
Harmful and Useful Microorganisms Presentationtahreemzahra82
 
Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |aasikanpl
 
Pests of safflower_Binomics_Identification_Dr.UPR.pdf
Pests of safflower_Binomics_Identification_Dr.UPR.pdfPests of safflower_Binomics_Identification_Dr.UPR.pdf
Pests of safflower_Binomics_Identification_Dr.UPR.pdfPirithiRaju
 
Grafana in space: Monitoring Japan's SLIM moon lander in real time
Grafana in space: Monitoring Japan's SLIM moon lander  in real timeGrafana in space: Monitoring Japan's SLIM moon lander  in real time
Grafana in space: Monitoring Japan's SLIM moon lander in real timeSatoshi NAKAHIRA
 
Scheme-of-Work-Science-Stage-4 cambridge science.docx
Scheme-of-Work-Science-Stage-4 cambridge science.docxScheme-of-Work-Science-Stage-4 cambridge science.docx
Scheme-of-Work-Science-Stage-4 cambridge science.docxyaramohamed343013
 
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCRCall Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCRlizamodels9
 
Pests of jatropha_Bionomics_identification_Dr.UPR.pdf
Pests of jatropha_Bionomics_identification_Dr.UPR.pdfPests of jatropha_Bionomics_identification_Dr.UPR.pdf
Pests of jatropha_Bionomics_identification_Dr.UPR.pdfPirithiRaju
 

Recently uploaded (20)

GenBio2 - Lesson 1 - Introduction to Genetics.pptx
GenBio2 - Lesson 1 - Introduction to Genetics.pptxGenBio2 - Lesson 1 - Introduction to Genetics.pptx
GenBio2 - Lesson 1 - Introduction to Genetics.pptx
 
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptxRESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
 
Forest laws, Indian forest laws, why they are important
Forest laws, Indian forest laws, why they are importantForest laws, Indian forest laws, why they are important
Forest laws, Indian forest laws, why they are important
 
Artificial Intelligence In Microbiology by Dr. Prince C P
Artificial Intelligence In Microbiology by Dr. Prince C PArtificial Intelligence In Microbiology by Dr. Prince C P
Artificial Intelligence In Microbiology by Dr. Prince C P
 
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxSOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
 
Evidences of Evolution General Biology 2
Evidences of Evolution General Biology 2Evidences of Evolution General Biology 2
Evidences of Evolution General Biology 2
 
TOPIC 8 Temperature and Heat.pdf physics
TOPIC 8 Temperature and Heat.pdf physicsTOPIC 8 Temperature and Heat.pdf physics
TOPIC 8 Temperature and Heat.pdf physics
 
Transposable elements in prokaryotes.ppt
Transposable elements in prokaryotes.pptTransposable elements in prokaryotes.ppt
Transposable elements in prokaryotes.ppt
 
OECD bibliometric indicators: Selected highlights, April 2024
OECD bibliometric indicators: Selected highlights, April 2024OECD bibliometric indicators: Selected highlights, April 2024
OECD bibliometric indicators: Selected highlights, April 2024
 
Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.
Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.
Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.
 
Pests of soyabean_Binomics_IdentificationDr.UPR.pdf
Pests of soyabean_Binomics_IdentificationDr.UPR.pdfPests of soyabean_Binomics_IdentificationDr.UPR.pdf
Pests of soyabean_Binomics_IdentificationDr.UPR.pdf
 
Solution chemistry, Moral and Normal solutions
Solution chemistry, Moral and Normal solutionsSolution chemistry, Moral and Normal solutions
Solution chemistry, Moral and Normal solutions
 
Twin's paradox experiment is a meassurement of the extra dimensions.pptx
Twin's paradox experiment is a meassurement of the extra dimensions.pptxTwin's paradox experiment is a meassurement of the extra dimensions.pptx
Twin's paradox experiment is a meassurement of the extra dimensions.pptx
 
Harmful and Useful Microorganisms Presentation
Harmful and Useful Microorganisms PresentationHarmful and Useful Microorganisms Presentation
Harmful and Useful Microorganisms Presentation
 
Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |
Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |
 
Pests of safflower_Binomics_Identification_Dr.UPR.pdf
Pests of safflower_Binomics_Identification_Dr.UPR.pdfPests of safflower_Binomics_Identification_Dr.UPR.pdf
Pests of safflower_Binomics_Identification_Dr.UPR.pdf
 
Grafana in space: Monitoring Japan's SLIM moon lander in real time
Grafana in space: Monitoring Japan's SLIM moon lander  in real timeGrafana in space: Monitoring Japan's SLIM moon lander  in real time
Grafana in space: Monitoring Japan's SLIM moon lander in real time
 
Scheme-of-Work-Science-Stage-4 cambridge science.docx
Scheme-of-Work-Science-Stage-4 cambridge science.docxScheme-of-Work-Science-Stage-4 cambridge science.docx
Scheme-of-Work-Science-Stage-4 cambridge science.docx
 
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCRCall Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
Call Girls In Nihal Vihar Delhi ❤️8860477959 Looking Escorts In 24/7 Delhi NCR
 
Pests of jatropha_Bionomics_identification_Dr.UPR.pdf
Pests of jatropha_Bionomics_identification_Dr.UPR.pdfPests of jatropha_Bionomics_identification_Dr.UPR.pdf
Pests of jatropha_Bionomics_identification_Dr.UPR.pdf
 

Deformable Part Models are Convolutional Neural Networks

  • 1. 1/26 Deformable Part Models are Convolutional Neural Networks Ross Girshick, Forrest Iandola, Trevor Darrell, Jitendra Malik Presentor: YANG Wei January 25, 2016
  • 2. 2/26 Outline 1 Introduction 2 DeepPyramid DPMs Feature pyramid front-end CNN Constructing an equivalent CNN from a DPM 3 Implementation details 4 Experiments
  • 3. 3/26 Outline 1 Introduction 2 DeepPyramid DPMs Feature pyramid front-end CNN Constructing an equivalent CNN from a DPM 3 Implementation details 4 Experiments
  • 4. 4/26 Deformable Part Models vs. Convolutional Neural Networks Deformable part models
  • 5. 4/26 Deformable Part Models vs. Convolutional Neural Networks Deformable part models Convolutional neural networks
  • 6. 5/26 Are DPMs and CNNs actually distinct? DPMs: graphical models CNNs: “black-box” non-linear classifiers This paper shows that any DPM can be formulated as an equivalent CNN, i.e., deformable part models are convolutional neural networks.
  • 7. 6/26 Outline 1 Introduction 2 DeepPyramid DPMs Feature pyramid front-end CNN Constructing an equivalent CNN from a DPM 3 Implementation details 4 Experiments
  • 8. 7/26 DeepPyramid DPMs Schematic model overview: “front-end CNN” + DPM-CNN input: image pyramid output: object detection scores
  • 9. 8/26 Feature pyramid front-end CNN front-end CNN: AlexNet (conv1-conv5). A CNN that maps an image pyramid to a feature pyramid AlexNet single-scale architecture
  • 10. 9/26 Constructing an equivalent CNN from a DPM A single-component DPM. mixture of components component = root filter + part filter
  • 11. 10/26 Inference with DPMs The matching process at one scale.
  • 12. 11/26 Architecture of DPM-CNN The unrolled detection algorithm of DPM generates a specific network with fixed length: 1 input: conv5 feature pyramid of front-end CNN
  • 13. 11/26 Architecture of DPM-CNN The unrolled detection algorithm of DPM generates a specific network with fixed length: 1 input: conv5 feature pyramid of front-end CNN 2 generate P+1 feature maps: 1 root filter and P part filters
  • 14. 11/26 Architecture of DPM-CNN The unrolled detection algorithm of DPM generates a specific network with fixed length: 1 input: conv5 feature pyramid of front-end CNN 2 generate P+1 feature maps: 1 root filter and P part filters 3 P part feature maps are fed into distance transform layer
  • 15. 11/26 Architecture of DPM-CNN The unrolled detection algorithm of DPM generates a specific network with fixed length: 1 input: conv5 feature pyramid of front-end CNN 2 generate P+1 feature maps: 1 root filter and P part filters 3 P part feature maps are fed into distance transform layer 4 root feature map are stacked (channel-wise concatenated) with the transformed part feature maps
  • 16. 11/26 Architecture of DPM-CNN The unrolled detection algorithm of DPM generates a specific network with fixed length: 1 input: conv5 feature pyramid of front-end CNN 2 generate P+1 feature maps: 1 root filter and P part filters 3 P part feature maps are fed into distance transform layer 4 root feature map are stacked (channel-wise concatenated) with the transformed part feature maps 5 The resulting P+1 channel feature map is convolved with an object geometry filter, which produces the output DPM score map for the input pyramid level
  • 17. 12/26 Architecture of DPM-CNN CNN equivalent to a single-component DPM.
  • 18. 13/26 Traditional distance transform Traditional distance transforms are defined for sets of points on a grid [FH05]. G: grid d(p−q): measure of distance between points p,q ∈ G B ⊆ G Then the distance transform of B on G DB(p) = min q∈B d(p−q) Distance transform (Euclidean distance)
  • 19. 14/26 Traditional distance transform DT can be also formulated as DB(p) = min q∈G (d(p−q)+1B(q)) where 1B(q) = 0, if q ∈ B, ∞, if q /∈ B. (1)
  • 20. 15/26 Generalized distance transform A generalization of distance transforms can be obtained by replacing the indicator function with some arbitrary function over the grid G Df (p) = min q∈G (d(p−q)+ f (q)) We can also define the generalized DT as maximization by letting f(q) = −f (q) Df (p) = max q∈G (f(q)−d(p−q))
  • 21. 16/26 Distance transform in DPM In DPM, after computing filter responses we transform the responses of the part filters to allow spatial uncertainty, Di(x,y) = max dx,dy (Ri(x+dx,y+dy)−wi ·φd(dx,dy)) where φd(dx,dy) = [dx,dy,dx2 ,,dy2 ] The value Di(x,y) is the maximum contribution of the part to the score of a root location that places the anchor of this part at position (x,y).
  • 22. 16/26 Distance transform in DPM In DPM, after computing filter responses we transform the responses of the part filters to allow spatial uncertainty, Di(x,y) = max dx,dy (Ri(x+dx,y+dy)−wi ·φd(dx,dy)) where φd(dx,dy) = [dx,dy,dx2 ,,dy2 ] The value Di(x,y) is the maximum contribution of the part to the score of a root location that places the anchor of this part at position (x,y). By letting p = (x,y), p−q = (dx,dy) and d(p−q) = w·φ(p−q), we can see that it is exactly in the form of distance transform.
  • 23. 17/26 Max pooling as distance transform Consider max pooling on f : G → R on a regular grid G. Let a window half-length as k, then max pooling can be defined as Mf (p) = max ∆p∈{−k,···,k} f(p+∆p) Max pooling can be expressed equivalently as distance transform: Mf (p) = max q∈G (f(q)−dmax(p−q)) where dmax(p−q) = 0, if (p−q) ∈ {−k,··· ,k}, ∞, otherwise . (2)
  • 24. 18/26 Generalize max pooling to distance transform pooling We can generalize max pooling to distance transform pooling: unlike max pooling, the distance transform of f at p is taken over the entire domain G rather than specifying a fixed pooling window a priori, the shape of the pooling region can be learned from the data.
  • 25. 18/26 Generalize max pooling to distance transform pooling We can generalize max pooling to distance transform pooling: unlike max pooling, the distance transform of f at p is taken over the entire domain G rather than specifying a fixed pooling window a priori, the shape of the pooling region can be learned from the data. The released code does not include the DT pooling layer. Please refer to [OW13] for more details.
  • 26. 19/26 Object geometry filters The root convolution map and the DT pooled part convolution maps are stacked into a
  • 27. 20/26 Combining mixture components with maxout CNN equivalent to a multi-component DPM. A multi-component DPM-CNN is composed of one DPM-CNN per component and a maxout [GWFM+13] layer that takes a max over component DPM-CNN outputs at each location.
  • 28. 21/26 Outline 1 Introduction 2 DeepPyramid DPMs Feature pyramid front-end CNN Constructing an equivalent CNN from a DPM 3 Implementation details 4 Experiments
  • 29. 22/26 Feature pyramid front-end CNN Implementation details pretrain on ILSVRC 2012 classification using Caffe use conv5 as output layer “same” convolution zero-pad each conv/pooling layer’s input with k/2 zeros on all sides (top, bottom, left and right) (x,y) in conv5 feature map has a receptive field centered on pixel (16x,16y) in the input image conv5 feature maps: stride: 16; receptive field: 163×163
  • 30. 23/26 Outline 1 Introduction 2 DeepPyramid DPMs Feature pyramid front-end CNN Constructing an equivalent CNN from a DPM 3 Implementation details 4 Experiments
  • 31. 24/26 Experiments Detection average precision (%) on VOC 2007 test. Column C shows the number of components and column P shows the number of parts per component.
  • 32. 25/26 Experiments HOG versus conv5 feature pyramids. In contrast to HOG features, conv5 features are more part-like and scale selective. Each conv5 pyramid shows 1 of 256 feature channels. The top two rows show a HOG feature pyramid and the face channel of a conv5 pyramid on the same input image.
  • 33. 26/26 References Pedro F Felzenszwalb and Daniel P Huttenlocher, Pictorial structures for object recognition, International Journal of Computer Vision 61 (2005), no. 1, 55–79. Ian J Goodfellow, David Warde-Farley, Mehdi Mirza, Aaron Courville, and Yoshua Bengio, Maxout networks, arXiv preprint arXiv:1302.4389 (2013). Wanli Ouyang and Xiaogang Wang, Joint deep learning for pedestrian detection, ICCV, IEEE, 2013, pp. 2056–2063.