local features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · matching local self-similarities across...

63
Local Features

Upload: others

Post on 10-May-2020

7 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Local Features

Page 2: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Correspondence Problem & Keypoint

• Correspondence: matching points, patches, edges, or regions across images

Page 3: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Keypoint Matching

K. Grauman, B. Leibe

Af Bf

A1

A2 A3

Tffd BA ),(

Page 4: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 5: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 6: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 7: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 8: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 9: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

DoG – Efficient Computation

• Computation in Gaussian scale pyramid

K. Grauman, B. Leibe

s

Original image 4

1

2s

Sampling with

step s4 =2

s

s

s

Page 10: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Find local maxima in position-scale space of Difference-of-Gaussian

K. Grauman, B. Leibe

)()( ss yyxx LL

s

s2

s3

s4

s5

List of (x, y, s)

Page 11: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Results: Difference-of-Gaussian

K. Grauman, B. Leibe

Page 12: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Challenges in Keypoints

• Scale change

• Rotation

• Occlusion

• Illumination

……

Page 13: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

SIFT

• Scale-space extrema detection

• Keypoint localization

• Orientation assignment

• Keypoint descriptor

Page 14: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Scale-space extrema detection

• Find the points, whose surrounding patches (with some scale) are distinctive

• An approximation to the scale-normalized Laplacian of Gaussian

Maxima and minima in a

3*3*3 neighborhood

Page 15: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Eliminating edge points

Page 16: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Orientation assignment

• Assign an orientation to each keypoint, the keypoint descriptor can be represented relative to this orientation and therefore achieve invariance to image rotation

• Compute magnitude and orientation on the Gaussian smoothed images

Page 17: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Orientation assignment

• A histogram is formed by quantizing the orientations into 36 bins;

• Peaks in the histogram correspond to the orientations of the patch;

• For the same scale and location, there could be multiple keypoints with different orientations;

Page 18: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

T. Tuytelaars, B. Leibe

Orientation normalization

• Compute orientation histogram

• Select dominant orientation

• Normalize: rotate to fixed orientation

0 2 p

Page 19: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

SIFT vector formation • Computed on rotated and scaled version of window

according to computed orientation & scale

– resample the window

• Based on gradients weighted by a Gaussian of variance half the window (for smooth falloff)

Page 20: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

SIFT vector formation • 4x4 array of gradient orientation histogram weighted by

magnitude

• 8 orientations x 4x4 array = 128 dimensions

• Motivation: some sensitivity to spatial layout, but not too much.

• Gaussian weight

showing only 2x2 here but is 4x4

Page 21: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

A result

Page 22: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 23: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 24: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 25: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 26: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 27: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 28: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 29: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 30: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 31: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 32: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 33: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 34: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 35: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 36: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 37: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 38: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Maximally Stable Extremal Regions [Matas ‘02]

• Select regions that stay stable over a large parameter range

K. Grauman, B. Leibe

Page 39: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Example Results: MSER

39 K. Grauman, B. Leibe

Page 40: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Self-similarity Descriptor

Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007

Page 41: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Self-similarity Descriptor

Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007

Page 42: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Self-similarity Descriptor

Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007

Page 43: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Local Descriptors

• Most features can be thought of as templates, histograms (counts), or combinations

• The ideal descriptor should be – Robust

– Distinctive

– Compact

– Efficient

• Most available descriptors focus on edge/gradient information – Capture texture information

– Color rarely used

K. Grauman, B. Leibe

Page 44: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

SIFT Repeatability

Lowe IJCV 2004

Page 45: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

SIFT Repeatability

Page 46: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Object Recognition

Computer Vision

Page 47: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 48: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 49: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 50: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 51: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 52: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 53: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 54: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 55: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 56: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can
Page 57: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

3D Object Recognition

Page 58: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

3D Shape Reconstruction

Page 59: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

RGBD Camera based object recognition

Page 60: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

3D Model based object recognition

Page 61: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

3D Model based object recognition

Page 62: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

3D Model based object recognition

Page 63: Local Features - khu.ac.krcvlab.khu.ac.kr/cvg10.pdf · Matching Local Self-Similarities across Images and Videos, Shechtman and Irani, 2007 . Local Descriptors •Most features can

Paper list

“Frustum PointNets for 3D Object Detection from RGB-D Data”

Charles R. Qi, Wei Liu, Chenxia Wu, Hao Su, Leonidas J. Guibas

“You Only Look Once: Unified, Real-Time Object Detection”

Joseph Redmon, Santosh Divvala, Ross Girshick, Ali Farhadi

“Squeeze-and-Excitation Networks”

Jie Hu, Li Shen, Gang Sun

“Adversarial Complementary Learning forWeakly Supervised Object

Localization”

Xiaolin Zhang, Yunchao Wei, Jiashi Feng, Yi Yang, Thomas Huang