HistogramBased Visual Object Recognition for the 2007 FourLegged - - PowerPoint PPT Presentation

histogram based visual object recognition for the 2007
SMART_READER_LITE
LIVE PREVIEW

HistogramBased Visual Object Recognition for the 2007 FourLegged - - PowerPoint PPT Presentation

HistogramBased Visual Object Recognition for the 2007 FourLegged RoboCup League Souzana Volioti and Michail G. Lagoudakis


slide-1
SLIDE 1

HistogramBased Visual Object Recognition for the 2007 FourLegged RoboCup League

Souzana Volioti and Michail G. Lagoudakis

  • !"
slide-2
SLIDE 2

!"#$

%&' ! (" )&*+

Outline

RoboCup Visual Object Recognition HistogramBased Object Recognition Results Conclusion

slide-3
SLIDE 3

!"#$

%&' ! (" )&*+

Outline

RoboCup

, $"-- , ." , " " ,

Visual Object Recognition HistogramBased Object Recognition Results Conclusion

slide-4
SLIDE 4

!"#$

%&' ! (" )&*+

RoboCup

RoboCup

, - , /0012/003).4- , $"- 5

Extensions

, $$"5 " , $"-6 $"- 4"$7

Vision

, 8#$%&% ''9 , ::" "; , ()*)#)+,(-

slide-5
SLIDE 5

!"#$

%&' ! (" )&*+

Where do we stand today?

slide-6
SLIDE 6

!"#$

%&' ! (" )&*+

RoboCup 2007 Soccer Leagues

Simulation League

, //+<// - :

SmallSize Robot League

, +<"-/

MiddleSize Robot League

, =+<="-

FourLegged Robot League

, 3+<3

Humanoid Robot League

, +<- "

slide-7
SLIDE 7

!"#$

%&' ! (" )&*+

Why RoboCup?

Research Challenges

, >?+ @ , : :>?# @ , ::>?A @ , : B>?- @ , :->? @ , :++>? @ , "":>?" @ , + +>? " @ , )>? ) @

Applicability

, "" + , "-B-"+<<<

slide-8
SLIDE 8

!"#$

%&' ! (" )&*+

RoboCup Participation

3/ 10C "A" $"- 1 =1

  • //

/3/ / 1 13 11 33 1/ Teams // '4- $"-/00C /0 DE $"- /00 1 ): $"- /000 /0 F"" $"-

  • *

$"- / E")")2"4- $"- 1 D " $"- 1 1C (D" $"- 3 1/ )4- $"- 1

  • $"- =

11 * $"- C Countries Place Event

slide-9
SLIDE 9

!"#$

%&' ! (" )&*+

Team

, ."5: , " 4"=

2007 Members

, . , +." , D) , B D , DD , D , "A!

Team Κουρήτες (Kouretes)

slide-10
SLIDE 10

!"#$

%&' ! (" )&*+

Where the team stands ...

Participation

, $"- =?&@ , $"- -C?E"( ("@ , $"- C?F$"@ , $"-? D5'F$G@

Distinctions

, C -?E"( (", $"- -C@ , -?F$", $"- C@ , -? D("G , $"- @ , 1 -? D("' , $"-@ , / -? D("F$, $"-@

Website

slide-11
SLIDE 11

!"#$

%&' ! (" )&*+

FourLegged League

Characteristics

, - , +"-- , ""-

History

, -: /00 , -:

Future

, :5 -" , ' " H $I , -:

slide-12
SLIDE 12

!"#$

%&' ! (" )&*+

Sony AIBO ERS7 Robots

slide-13
SLIDE 13

!"#$

%&' ! (" )&*+

Sony AIBO ERS7 Robots

slide-14
SLIDE 14

!"#$

%&' ! (" )&*+

2007 FourLegged League Field

slide-15
SLIDE 15

!"#$

%&' ! (" )&*+

2007 FourLegged League Game

slide-16
SLIDE 16

!"#$

%&' ! (" )&*+

Kouretes in Action

slide-17
SLIDE 17

!"#$

%&' ! (" )&*+

FourLegged League Challenges

Motion

, 51 - , 5 )):

Vision

, #5:"8"9 , 51

Localization

, " +: )--"

Behavior

, +-

slide-18
SLIDE 18

!"#$

%&' ! (" )&*+

Outline

RoboCup Visual Object Recognition

, -- , - , #

HistogramBased Object Recognition Results Conclusion

slide-19
SLIDE 19

!"#$

%&' ! (" )&*+

Image Preprocessing

Raw Camera Image

, 7J*! /=-B1- , +:5=<0 : 3< , 1 5 - ) , ++:53 : /C

Image Correction

, +

Color Segmentation

, $"-5)":: , -B , + ?C@

slide-20
SLIDE 20

!"#$

%&' ! (" )&*+

Problem Statement

Recognition

, A#5 , #-+: "+:

Estimation

, A #

Transformation

, -: - :

slide-21
SLIDE 21

!"#$

%&' ! (" )&*+

Field Objects: Goals

Goals

, 5)": , 50 1 1 , " C

slide-22
SLIDE 22

!"#$

%&' ! (" )&*+

Field Objects: Beacons

Beacons

, 5:2:2)":2)"2: , 53HI/H I , C

slide-23
SLIDE 23

!"#$

%&' ! (" )&*+

Field Objects: Ball

Ball

, 5 , 5/ , - +"

slide-24
SLIDE 24

!"#$

%&' ! (" )&*+

Outline

RoboCup Visual Object Recognition HistogramBased Object Recognition

, - , , ,

Related Work Results

slide-25
SLIDE 25

!"#$

%&' ! (" )&*+

Basic Operations

Construction of Horizon Line

, -# -

Construction of Scan Lines

, A2+-2-- "A

Identification of Large Parts

, " -

Construction of Density Histograms

, -++:

Estimation of Distance and Angle

, " -

slide-26
SLIDE 26

!"#$

%&' ! (" )&*+

Example: SkyBlue Goal Scanning

Raw Image Horizon/Scan Lines Large Parts/Histograms Indications/Perception

K1<C K1

slide-27
SLIDE 27

!"#$

%&' ! (" )&*+

Goal Recognition (1)

Construction of Horizontal Scan Lines

, -A , 3"

slide-28
SLIDE 28

!"#$

%&' ! (" )&*+

Goal Recognition (2)

Identification of Horizontal Large Parts

, 5NG " 2G " , + -B"

slide-29
SLIDE 29

!"#$

%&' ! (" )&*+

Goal Recognition (3)

Horizontal histogram

, - , A ++: , ""

, -)

slide-30
SLIDE 30

!"#$

%&' ! (" )&*+

Goal Recognition (4)

Vertical scanning, large parts, histograms

slide-31
SLIDE 31

!"#$

%&' ! (" )&*+

Goal Recognition (5)

Scanning constraint

, +:1 " A

Filtering

, #-"

Confidence filter

, -"+

Ratio filter

, A2+""

Distance filter

, +

slide-32
SLIDE 32

!"#$

%&' ! (" )&*+

Beacon Recognition (1)

Horizontal scanning, large parts, histograms

slide-33
SLIDE 33

!"#$

%&' ! (" )&*+

Beacon Recognition (2)

Vertical scanning, large parts, histograms

slide-34
SLIDE 34

!"#$

%&' ! (" )&*+

Beacon Recognition (3)

Scanning constraint

, +:1 " A

Filtering

, #-"

Proximity filter

, +-)" #

Ratio filter

, A2+""

Distance filter

, +

slide-35
SLIDE 35

!"#$

%&' ! (" )&*+

Ball Recognition (1)

Horizontal scanning, large parts, histogram

slide-36
SLIDE 36

!"#$

%&' ! (" )&*+

Ball Recognition (2)

Vertical scanning, large parts, histogram

slide-37
SLIDE 37

!"#$

%&' ! (" )&*+

Ball Recognition (3)

Ball Center and Radius

, --)A + , - , 1--- , 5 -- " , "5 : ---

slide-38
SLIDE 38

!"#$

%&' ! (" )&*+

Ball Recognition (4)

Scanning constraint

, +:A

Filtering

, #-"

Horizon filter

, ":A

Distance filter

, +

slide-39
SLIDE 39

!"#$

%&' ! (" )&*+

Outline

RoboCup Visual Object Recognition HistogramBased Object Recognition Results

, +"- , -&3 DL$= , --

Conclusion

slide-40
SLIDE 40

!"#$

%&' ! (" )&*+

Evaluation Methodology

Evaluation

, 5"":- , 51 "$"--C , 5M/-

Competitors

, &35# &3 , DL$=5# DL$( =

Comparison

, - , """"2

slide-41
SLIDE 41

!"#$

%&' ! (" )&*+

Evaluation Results

slide-42
SLIDE 42

!"#$

%&' ! (" )&*+

Perception Differences

GT 2004 SPQR 2006 Proposed

slide-43
SLIDE 43

!"#$

%&' ! (" )&*+

Perception Differences

GT 2004 SPQR 2006 Proposed

slide-44
SLIDE 44

!"#$

%&' ! (" )&*+

Outline

RoboCup Visual Object Recognition HistogramBased Object Recognition Results Conclusion

slide-45
SLIDE 45

!"#$

%&' ! (" )&*+

Conclusion

Summary

, "--A , -" - , --+"--

Application

, " ."$"-C

Lessons

, -+"--- , -+<- + , "-:+ "+

slide-46
SLIDE 46

!"#$

%&' ! (" )&*+

Acknowledgements

Technical University of Crete, Chania

, "+ , "---- , $"-=2C2-C

Parish of Panagitsa, Kounoupidiana

, $+<E< 7B ) , +"- +- , 4"C, FC

European Commission

, F"$$330 , -"--

slide-47
SLIDE 47

Thank you!