semantic texton forests for image categorization and segmentation

24
Semantic Texton Forests for Image Categorization and Segmentation We would like to thank Amnon Drory for this deck ההההה: ההההה הההההה ההה ההההה ההההה ההההה ההה הה הההההה / הה ההההה ההההה.

Upload: nikkos

Post on 23-Feb-2016

57 views

Category:

Documents


1 download

DESCRIPTION

Semantic Texton Forests for Image Categorization and Segmentation. We would like to thank Amnon Drory for this deck. הבהרה: החומר המחייב הוא החומר הנלמד בכיתה ולא זה המופיע / לא מופיע במצגת. . Semantic Texton Forests. Input: - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Semantic Texton Forests for Image Categorization and Segmentation

Semantic Texton Forests for Image Categorization and

Segmentation

We would like to thank Amnon Drory for this deck

הבהרה: מופיע / לא המופיע זה ולא בכיתה הנלמד החומר הוא המחייב החומר

במצגת.

Page 2: Semantic Texton Forests for Image Categorization and Segmentation

Semantic Texton Forests Input: 1. Training: Images with pixel level ground truth

classification

MSRC 21 Database

Page 3: Semantic Texton Forests for Image Categorization and Segmentation

Semantic Texton Forests Input: 1. Training: Images

with pixel level ground truth classification.

2. Testing: Images

Output: A classification of each pixel in the test images to an object class.

Page 4: Semantic Texton Forests for Image Categorization and Segmentation

Main Mathematical Tools

1. Conditional Random Fields (CRF)

2. Textons ( Convolution with Filter Banks )

3. K-Means

4. Joint Boost

Randomized Decision Forests

Page 5: Semantic Texton Forests for Image Categorization and Segmentation

Decision Trees

Page 6: Semantic Texton Forests for Image Categorization and Segmentation

Decision Trees

Page 7: Semantic Texton Forests for Image Categorization and Segmentation

Decision Rules (Split Functions)

1. Choose one or two pixels near pixel of interest

2. Calculate simple function of their values:- Difference- Sum- Absolute Difference- Or just the pixel

valuere To a hresholdMotivation: very small number of computer operations. Easy to implement on GPU.

Page 8: Semantic Texton Forests for Image Categorization and Segmentation

Training the treeFor each node:1.Randomly generate a few decision rules2.Choose the one that maximally improves the ability of the tree to separate between classes. (E(I) – entropy)

Stop when tree reached pre-defined depth , or when no further improvement in classification can be achieved.

Page 9: Semantic Texton Forests for Image Categorization and Segmentation

Decision Forests

For added strength, create several trees instead of just one. Each tree is trained using a different subset of the training data.

Page 10: Semantic Texton Forests for Image Categorization and Segmentation

Classifying at test time

For each pixel in the test image:Apply the segmentation forest – marking a path in each tree (yellow).Each leaf is associated with a histogram of classes.Average the histograms from all tree, achieving a vector of probabilities for this pixel belonging to each class: 0.0

10.23

0.05

0.04

0.001

Page 11: Semantic Texton Forests for Image Categorization and Segmentation

Classifying at test time

The probability vectors derived from the Decision Forests can be used to classify pixels to classes, by assigning to each pixel the label that is most likely. The results are very noisy.

0.01

0.23

0.05

0.04

0.001

Page 12: Semantic Texton Forests for Image Categorization and Segmentation

Paths on trees represent Texture1

2 34 5 6 7

8 9 1011

17

8485 86

8791

The texture of an area around a pixel can be represented by a vector comprised of all the

nodes in the decision forest that belong to paths traversed when applying the forest to

this pixel. In the above example, this would be the vector :

(1 ,3 ,6 ,10 ,17 , … ,84 ,85 ,87 ,91) This vector is called a Semantic Texton.

Page 13: Semantic Texton Forests for Image Categorization and Segmentation

Example of Semantic Textons

A visualization of leaf nodes fromone tree (distance d = 21 pixels). Each patch is

the average of all patches in the training images assigned to a particular leaf node. Features

evident include color, horizontal, vertical and diagonal edges, blobs, ridges and corners .

Page 14: Semantic Texton Forests for Image Categorization and Segmentation

Example of Semantic Textons

Page 15: Semantic Texton Forests for Image Categorization and Segmentation

Second Randomized Decision Forest

• The split functions at the nodes are based on the Texture-Layout filter:

• Two types of Texture-Layout filters are used:1. Count the occurrence of a certain node in the

semantic textons of pixels in a rectangle.( , )

Page 16: Semantic Texton Forests for Image Categorization and Segmentation

Second Randomized Decision Forest

• The split functions at the nodes are based on the Texture-Layout filter:

• Two types of Texture-Layout filters are used:1. Count the occurrence of a certain node N in the

semantic textons of pixels in a rectangle.2. Sum the probabilities of belonging to a certain class K

for all pixels in a rectangle. This is semantic context.

( , )

Page 17: Semantic Texton Forests for Image Categorization and Segmentation

STF - Results

Overall Accuracy: 72%Though less aesthetic, these results are quantitatively

almost as good as those of TextonBoost.

Page 18: Semantic Texton Forests for Image Categorization and Segmentation

STF - Summary

• To speed up calculation, this algorithm uses Radomized Decision Forests instead of other mathemaical tools used in TextonBoost.

• One RDF is used to calculate texture.• A second RDF is used to classifiy pixels to object

classes. • The results are almost as good (quantitatively) as

those of TextonBoost.• The algorithm is much quicker than TextonBoost.

Page 19: Semantic Texton Forests for Image Categorization and Segmentation

Real-Time Human Pose Recognition in Parts from

Single Depth Images

J. Shotton, A. Fitzgibbon, M. Cook, T. Sharp, M. Finocchio, R. Moore,

A. Kipman, A. Blake

Page 20: Semantic Texton Forests for Image Categorization and Segmentation

Overview

Page 21: Semantic Texton Forests for Image Categorization and Segmentation

Classification• Classify each pixel to one of 31 body part classes:

– Left Upper / Right Upper / Left Lower / Right Lower head– neck– Left / Right shoulder– Left Upper / Right Upper / Left Lower / Right Lower arm– Left /Right elbow– Left /Right wrist– Left /Right hand– Left Upper / Right Upper / Left Lower / Right Lower torso– Left Upper / Right Upper / Left Lower / Right Lower leg– Left /Right knee– Left /Right ankle– Left /Right foot

Page 22: Semantic Texton Forests for Image Categorization and Segmentation

Randomized Decision Forests

• The Randomized Decision Forests are very deep (depth = 20).

• => A very strong ability to classify correctly on the Training Set.

• => A risk of over fitting. This risk is averted by using a very (very) large training set, containing examples of many poses we wish to recognize. Most of this training set is artificially generated.

Page 23: Semantic Texton Forests for Image Categorization and Segmentation

Decision Rules

Page 24: Semantic Texton Forests for Image Categorization and Segmentation

Classification → body part locations

• Separately for each of 31 Body Part Classes:– Create Probability map– Project to 3D space– Find modes ( using mean-shift )– The modes (with a little post processing ) are the

suggested body part locations (the output of this algorithm).