Advanced Computer Vision

beeuppityΤεχνίτη Νοημοσύνη και Ρομποτική

19 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

116 εμφανίσεις

Roger S. Gaborski

1

Advanced Computer Vision


Lecture
03

Roger S. Gaborski

Video Lecture


http://videolectures.net/
nips09_torralba_uv
s


Paper:
VLFeat
-
An open and portable library of computer vision
algorithms


http://
vision.ucla.edu
/papers/vedaldiF10.pdf


Roger S. Gaborski

2

Roger S. Gaborski

3

Object Recognition


Issues:


Viewpoint


Scale


Deformable vs. rigid


Clutter


Occlusion


Intra class variability

Roger S. Gaborski

4

Goal


Locate all instances of automobiles in a
cluttered scene

Roger S. Gaborski

5

Acknowledgements


Students (Thesis in RIT Library):


Tim Lebo


Dan Clark



Images used in presentation:


ETHZ Database, UIUC Database

Roger S. Gaborski

6

Object Recognition Approaches


For specific object class:


Holistic


Model whole object


Parts based


Simple parts


Geometric relationship
information


We could use a similar approach to match
patches representing different image
categories (‘sand’ patches located on
lower half of beach scenes)



Roger S. Gaborski

7

Training Images and Segmentation

Roger S. Gaborski

8

Implicit Shape Model


Patches


local appearance prototypes


Spatial relationship


where the patch can be
found on the object


For a given class w:

ISM(w) = (I
w

,P
w

)

where I
w

is the codebook containing the patches

and P
w

is the probability distribution that describes

where the patch is found on the object



How do we find

interesting


patches?

Roger S. Gaborski

9

Harris Point Operator

Roger S. Gaborski

10

Harris Points

Roger S. Gaborski

11

Segmented Training Mask

Segmented mask ensures only patches containing valid car regions are selected


A corresponding segmentation patch is also extracted

Roger S. Gaborski

12

Selected Patches

Roger S. Gaborski

13

How is spatial information
represented?


Estimate the center of the object using the
centroid of the segmentation mask


Displacement between:


Center of patch


Centroid of segmentation mask

Roger S. Gaborski

14

Individual Patch and Displacement
Information

Roger S. Gaborski

15

Typical Training Example

Roger S. Gaborski

16

Typical Training Example

Roger S. Gaborski

17

Extracted Training Patches

Roger S. Gaborski

18

Cluster Patches


Many patches will be visually similar


Normalized Grayscale Correlation is used
to cluster patches


All patches within a certain neighborhood
defined by the NGC are grouped together


The representative patch is determined by
mean of the patches


The geometric information for each patch in
the cluster is assigned to the representative
patch

Roger S. Gaborski

19

Patches

Roger S. Gaborski

20

Wheel Patch Example

Roger S. Gaborski

21

Clusters

Opportunity for better clustering method

Roger S. Gaborski

22

Clusters

Roger S. Gaborski

23

Roger S. Gaborski

24

Object Detection


Harris point operator to find interesting
points


Extract patches


Match extracted patches with model
patches


Spatial information predicts center of
object


Create voting space

Roger S. Gaborski

25

Ideal Voting Space Example

Roger S. Gaborski

26

Multiple Votes

Multiple geometric interpretations

Roger S. Gaborski

27

Resolving False Detections

Roger S. Gaborski

28

Localization: Find Corners

Roger S. Gaborski

29

Localization: Model Matching

Roger S. Gaborski

30

Localization: Find Corners

Roger S. Gaborski

31

Model Matching

Roger S. Gaborski

32

Spatial Activation

(Hough Space)

9000 different locations

Roger S. Gaborski

33

Hypothesis Candidates

16 candidate locations

Roger S. Gaborski

34

Hypothesis Candidates

Roger S. Gaborski

35

References

SEE RESOURCES ON COURSE WEB PAGE:



Timothy Lebo and Roger Gaborski,

A Shape model with Coactivation
Networks for Recognition and Segmentation,


Eighth International conference
on Signal and image Processing, Honolulu, HI. August 2006.


Timothy Lebo,

Guiding Object Recognition: A Shape Model with Co
-
activation
Networks,


MS Thesis, RIT, 2005.


Daniel Clark,

Object Detection and Tracking using a Parts Based Approach,


MS Thesis, RIT, 2005.




Roger S. Gaborski

36

References


Bastian Leibe, Ales Leonardis, and Bernt Schiele,

Combined object
categorization and segmentation with an implicit shape model,


ECCV

04
Workshop onStatistical Learning in Computer Vision, May 2004.


Shivani Agarwal, Aatif Awan, and Dan Roth,

Learning to detect objects in
images via a sparse, part
-
based representation,


IEEE Transactions on
Pattern Analysis and Machine Intelligence, 26(11):1475

1490, 2004.




Roger S. Gaborski

37

Roger S. Gaborski

38

Voting Space


Roger S. Gaborski

39

Roger S. Gaborski

40

Model Patches Selected


Roger S. Gaborski

41

True Object Patches


Roger S. Gaborski

42

Identified Objects


Research Topics


Scene Categories


MATLAB TOOLKIT: http
://
www.vlfeat.org
/

Roger S. Gaborski

43

Roger S. Gaborski

44

Reference Examples

Recognizing Indoor Scenes

Ariadna

Quattoni

and Antonio
Torralba

(http://
people.csail.mit.edu
/
torralba
/publications/
indoor.pdf
)


Building the gist of a scene: the role of global image

features in recognition

Aude

Oliva

and Antonio
Torralba


Objects as Attributes for Scene
Classi¯cation

Li
-
Jia

Li,
Hao

Su,
Yongwhan

Lim, Li
Fei
-
Fei

(
http://vision.stanford.edu/documents/LiSuLimFeiFei_ECCV2010.pdf
)


See references for each paper



Roger S. Gaborski

45


Modeling the shape of the scene: a holistic representation of the
spatial envelope

http://people.csail.mit.edu/torralba/code/spatialenvelope/



SIFT flow: dense correspondence across difference scenes

http://people.csail.mit.edu/celiu/ECCV2008/


Roger S. Gaborski

46


Learning
and Recognizing Visual Object
Categories


http://
www.youtube.com
/
watch?v
=w2C
-
WffS
-
AE&feature=
bf_prev&list
=PL9415E136FBEE3016&lf=
results_m
ain

Deep Learning:
Visual Perception with Deep Learning

http
://
www.youtube.com
/
watch?v
=3boKlkPBckA

Typical Images

Roger S. Gaborski

47

coast_cdmc976.jpg


coast_n708004
.
jpg

coast_n384026.jpg


coast_n739047
.jpg


forest_for142.jpg


forest_for42
.jpg

forest_for38.jpg


forest_for58
.jpg


tallbuilding_a487092.jpg

tallbuilding_a803053.jpg

tallbuilding_art1350.jpg

tallbuilding_art1506.jpg


http:/
/
labelme.csail.mit.edu
/Images/spatial_envelope_256x256_static_8outdoorcategories
/

Coast Images

Roger S. Gaborski

48

Forest

Roger S. Gaborski

49

Tall Buildings

Roger S. Gaborski

50

Canny, I1
-
I4

Roger S. Gaborski

51

Canny, I5
-
I9

Roger S. Gaborski

52

Canny, I10
-
I12

Roger S. Gaborski

53

Coast Images

Roger S. Gaborski

54

Coast


Red Data

Roger S. Gaborski

55

Coast


Red Histogram

Roger S. Gaborski

56

H
W#3 Assignment


Work individually


Select a paper relevant to a project you
are considering (one of 3 topics)


Prepare a review of the paper


Due 12/
13


Make
a 10
minute in class presentation


Email your slides to the course account by
6pm 12/13


Roger S. Gaborski

57