Week1 Report

15
Week 1 Report: +Different of Gaussian (DoG) + SIFT: Use to detect keypoint Describe the features of keypoint.

Transcript of Week1 Report

Page 1: Week1 Report

Week 1 Report: +Different of Gaussian (DoG) + SIFT:

Use to detect keypoint Describe the features of keypoint.

Page 2: Week1 Report

Bag of features: • Detection and description of image patches • Assigning patch descriptors to a set of predetermined clusters (a vocabulary) with a vector quantization algorithm • Constructing a bag of keypoints, which counts the number of patches as-signed to each cluster • Applying a multi-class classifier, treating the bag of keypoints as the feature vector, and thus determine which category or categories to assign to the image.

Page 3: Week1 Report

Matlab program: Use Vlfeat library: SIFT detector (plot descriptor)

(a) (b)

(c) (d)

SIFT detector and descriptor: (a) Mud, (b) ripple, (c)posidonie, (d) sand_posi.

Page 4: Week1 Report

Bag of features Matlab results: Demo program: Also use Vlfleats library, dense SIFT descriptor, Support vector machine classifier ( linear SMV). Use caltech101 data with 5 classes: Accordion, airplane, budha, ant and camera 15 training images, 15 data images. Confusion matrix:

Confusion matrix (92.00 % accuracy)

accordion airplanes ant budha camera

accordion

airplanes

ant

budha

camera

Page 5: Week1 Report

GMM: KMeans intialization

GMM: Gaussian mixture - kmeans init

GMM: Gaussian mixture - random init

Page 6: Week1 Report

Gaussian Mixture model: The Gaussian Mixture Model (GMM)

The log-likelihood of this model:

EM algorithm:

Repeat until Loglikelihood convergence.

Page 7: Week1 Report

Results: Training with 20 sona images.

(a) (b) (c)

1 2 3 4 50

200

400

600

800

1000

1200

1400

1600

1800

2000

1 2 3 4 50

200

400

600

800

1000

1200

1400

1600

1800

1 2 3 4 50

200

400

600

800

1000

1200

1400

1600

1800

Page 8: Week1 Report

(d) (e)

Keypoint classification of images: (a) Sona1, (b) Sona2, (c) Sona3, (d) Sona4, (e) Sona5

1 2 3 4 50

200

400

600

800

1000

1200

1400

1600

1800

1 2 3 4 50

200

400

600

800

1000

1200

1400

Page 9: Week1 Report

Week 3: Strauss Point Process:

0 10 20 30 40 50 60 70 80 90 1000

1

2

3

4

5

6

7

0 10 20 30 40 50 60 70 80 90 100-2940

-2920

-2900

-2880

-2860

-2840

-2820

-2800

-2780

-2760

-2740

Page 10: Week1 Report

Gamma Predict Maximum Pseudo Likelihood

Page 11: Week1 Report

Choose r=6, Gamma= 0.614

Condition Intensity function Kest

Model diagnostics (raw residuals) Diagnostics available: four-panel plot mark plot smoothed residual field x cumulative residuals y cumulative residuals sum of all residuals sum of raw residuals in clipped window = -1.575e-11 8

Page 12: Week1 Report
Page 13: Week1 Report

Poisson

Strauss r=6 StraussHardcore r=6

Softcore(kappa=0.5)

PairPiece (r=5)

DiggleGratton

(0.05,0.2)

LennardJones

Sona01_01 -2181.348 -2150.018 -2149.969 -2179.729 -2148.034 -2179.87 -2179.083

Sona01_21 -1780.236 -1765.172 -1764.972 -1780.183 -1773.054 -1780.236 -1774.519

Sona01_40 -2285.435 -2219.694 -2218.69 -2283.463 -2235.879 -2285.435 -2283.811

Sona02_01 -1747.328 -1728.662 -1728.614 -1747.326 -1737.707 -1746.569 -1740.42

Sona02_21 -1714.287 -1695.617 -1695.396 -1714.108 -1700.734 -1714.04 -1712.54

Sona02_40 -2128.879 -2107.925 -2107.925 -2122.054 -2120.438 -2122.48 -2122.265

Sona03_01 -1110.386 -1107.001 -1106.953 -1108.335 -1109.643 -1110.386 -1105.97

Sona03_21 -882.6423 -881.5504 -881.193 -882.559 -882.2873 -882.5832 -880.6295

Sona03_40 -1055.875 -1053.122 -1052.917 -1055.874 -1055.841 -1055.875 -1050.915

Sona04_01 -582.965 -582.7899 -582.7004 -582.6918 -582.4404 -582.9356 -33789152

Sona04_21 -1019.234 -1018.985 -1018.923 -1011.754 -1018.683 -1019.164 -1011.711

Sona04_40 -1217.897 -1213.966 -1213.544 -1217.783 -1216.747 -1217.897 -1217.039

Sona05_01 -2264.706 -2220.523 -2220.523 -2261.285 -2238.694 -2261.417 -2261.026

Sona05_21 -1697.715 -1691.061 -1691.001 -1696.457 -1694.802 -1695.566 -1693.013

Sona05_40 -1921.357 -1900.423 -1900.423 -1915.615 -1905.489 -1915.436 -1915.371

Time of maximum

0 3 8 0 1 4

Page 14: Week1 Report

P-value of homogenous StraussHard(r=roptimum) with no rbord (edge effect)

Image Optimum r p-value

Sona01_01 5.3 0.0003015077

Sona01_21 6.2 0.008560914

Sona01_40 6.4 7.128579e-08

Sona02_01 6.4 0.001921654

Sona02_21 6.4 0.002268519

Sona02_40 5.9 0.01058475

Sona03_01 6.1 0.2263023

Sona3_21 4.4 0.3377286

Sona03_40 6.4 0.1437486

Sona04_01 8.8 0.3081564

Sona04_21 7 0.3087066

Sona04_40 6 0.2521565

Sona05_01 5.9 3.354012e-05

Sona05_21 6.4 0.04243896

Sona05_40 6.7 0.000384445

P-value of inhomogenous StraussHard(r=roptimum) model with rbord=9.5

Image Optimum r p-value sum of raw residuals

LennardJones

Sona01_01 5.4 1.629268e-05 -2.693e-08 0.7953043

Sona01_21 6.1 0.003474408 -1.401e-09 0.1619573

Sona01_40 6.4 1.541811e-08 -9.213e-09 0.9114976

Sona01_15 6.3 0.0001993449 -5.264e-11 0.2343573

Sona01_30 6.6 6.290544e-07 -5.187e-10 0.2132667

Sona02_01 6.4 5.674017e-05 -2.522e-08 0.6585665

Sona02_21 6.5 5.357882e-06 -5.651e-08 0.5456646

Sona02_40 6.1 0.001032961 -1.626e-11 0.1234234

Sona02_15 5.9 1.368404e-08 -6.802e-08 0.4565464

Sona02_30 6.1 3.0307e-06 3.0307e-06 0.1619573

Sona03_01 6.7 0.006028275 -3.703e-10 0.3838797

Sona3_21 2.4 0.1544988 -2.091e-07 0.8038365

Sona03_40 6.6 0.05494958 -6.225e-07 0.6917028

Sona03_15 5.4 6.30801e-09 2.119e-12 2.377465e-11

Sona03_30 5.4 0.1892866 -5.623e-09 0.08558589

Sona04_01 8.8 0.06928745 -2.091e-07 0.1877029

Sona04_21 6.9 0.3051947 -4.323e-07 0.06520478

Sona04_40 7 0.03310093 -2.656e-08 0.8916772

Sona04_15 9.5 0.03410365 -4.198e-08 0.3112048

Sona04_30 8.3 0.005665654 -5.691e-10 0.4009809

Sona05_01 6.7 1.09733e-06 -9.549e-11 0.7435345

Sona05_21 6.4 0.009255479 -5.651e-08 0.2342342

Sona05_40 6.7 2.782162e-06 -1.626e-11 0.2343241

Sona05_15 6.3 0.0127437 -6.802e-08 0.1132332

Sona05_30 6.7 6.612725e-06 3.0307e-06 0.6546568

Page 15: Week1 Report

QQplot, Strauss model, Edge correction, Sona01_01