Anda di halaman 1dari 32

Computer)Vision:)Course)Introduction) and)Basics)

Lecture'1'

1"

Lecture)Etiquette)
On#Ques(ons# ! There"is"no"such"thing"as"a"stupid"ques5on" ! Unless"we"ques5on,"we"will"not"learn:"Who$ques)ons$ much,$shall$learn$much,$and$retain$much.$Francis"Bacon" ! Why"are"ques5ons"so"important?"Because"the"process"of" scholarship"is"more"important"than"the"facts"themselves" ! When"you"ques5on,"you"are"a"selfEmo5vated"learner""a" researcher" ! Learning"is"a"dialogue,"not"transmission"and"recep5on" ! Your"ques5ons"will"help"your"classEmates" ! Ques5ons"are"an"important"feedback"mechanism"
Computer'Vision''Fall'2011' 2"

Introduction)
!

Sight"is"our"primary"sensa5on"
! !

80%"of"our"rst"12"years"of"learning"is" through"vision" 40%"of"the"brain"is"dedicated"to"visual" processing"

Human"Experience" ! Vision"in"language"
!
! ! !

Seeing"is"believing" Jinnah"was"a"man"of"vision"" Love"is"blind"

Computer'Vision''Fall'2011'

Sight)and)Vision)in)Urdu)Poetry)
&'&( '" )* +" !#" $% ,-$ &-% '" !"#$ !" !#" $% ,-$ ,-$ &'./ 012 "!' $- ,!2 +"#3 $% 2'" +" !#" -4 *'.3 ,'!0#. $6'0 %&# '() +5 ,!&" *'*/. - $7-" +8-# ,!$% ()*# !&2 ,!2 6'42 #'.) 3 %&#9&:0%2 0(!(< +' !=." ,!2 6'!% %!.4 !&2 ,!$ +$& >?0 +*4# ,-&'6$ $"

Computer'Vision''Fall'2011'

What)does)it)mean)to)see?)
! ! !

To"know"what"is"where"by"looking"" "EE"Aristotle" Computer"Vision"Scien5st:"How"do"we" build"a"machine"to"do"that?" Scien5cally"unsa5sfactory"


! !

SKY

MOUNTAIN

What"is"what?" What"is"where?"
FOLIAGE

Ongoing"inves5ga5on"
! !

Iden5ca5on" Localiza5on"

Computer'Vision''Fall'2011'

WALL

MAN

FENCE

What)is)Computer)Vision)
!

The"goal"of"Computer"Vision"is"to"make"useful"decisions" about"real"physical"objects"and"scenes"based"on"sensed" images"


Symbolic Decision OUT

Image IN

Image Processing

Image OUT

Image IN

Computer Vision

Symbolic Info IN

Computer Graphics

Image OUT

Computer'Vision''Fall'2011'

What)is)computer)vision?)
Optics Geometry Statistics Machine Learning Neurobiology

Artificial Intelligence

Computer Vision

Image Processing

Robotic Control

Surveillance

Virtual Reality

Medical Imaging

Document Analysis

Computer'Vision''Fall'2011'

Why)is)computer)vision)hard?)
! !

Computers"are"good"at"numerical"processing" Humans"are"good"at"perceptual"processing" We"want"to"use"a"computer"to"mimic"human"percep5on" which"is"complex"to"understand"

Computer'Vision''Fall'2011'

The)Complexity)of)Perception)

The)Complexity)of)Perception)

Computer'Vision''Fall'2011'

By Edward Adelson, 1995

Perception)

Computer'Vision''Fall'2011'

Perception)

Computer'Vision''Fall'2011'

Perception)

Ref: Light and Vision: LIFE Science Library Computer'Vision''Fall'2011'

What)is)this?)

Computer'Vision''Fall'2011'

Computer'Vision''Fall'2011'

The)Complexity)of)Perception)

Computer'Vision''Fall'2011'

The)complexity)of)images)

It"seems"like"a"hopeless"task"to"be"able"to"write"a" program"to"interpret"these"images"
Computer'Vision''Fall'2011'

Writing)Programs)that)See)
An'Example'

Motivation)
Humans"have"highly"sophis5cated"capabili5es"of"sensing" the"environment,"interpre5ng"it"and"taking"ac5ons" accordingly" ! Human"Percep5on:"the$process$of$a8aining$ awareness$or$understanding$of$sensory$informa)on" ! Can"machines"have"similar"capabili5es?"
!
!

Can"we"write"algorithms"for"percep5on?"

Computer'Vision''Fall'2011'

Motivation)
Machines"Compute$$ $Humans"Perceive"

Can"percep)on"arise"from"computa)on?"

Computer'Vision''Fall'2011'

Gary)Kasparov)vs.)Deep)Blue)(1997))

! ! ! ! ! ! !

Can"evaluate"about"3"posi5ons"per" second" Large"chess"knowledge,"low" computa5on"ability" Uses"a"lot"of"intui5on"and"feeling" Learns"and"adapts"very"quickly"from" his"mistakes" Can"get"bored,"fa5gued,"loss"of" concentra5on" Highly"intelligent" Selec5ve"searching"of"posi5ons"
Computer'Vision''Fall'2011'

! ! ! ! ! ! !

Can"evaluate"about"200,000,000" posi5ons"per"second" Small"chess"knowledge,"high" computa5on"ability" Uses"only"computa5ons" Not"a"learning"system"(machine" learning"algorithms"were"not"used)" Not"aected"by"feelings"such"as" Kasparovs"stare" Dumber"than"a"2"year"old" BruteEforce"evalua5on"of"all"moves"

hhp://www.research.ibm.com/deepblue/"

Motivation)
Machines"Compute$$ $Humans"Perceive"

Can"percep)on"arise"from"computa)on?"

Computer'Vision''Fall'2011'

Why)is)Perception)DifPicult)for)) Computational)Machines?)
!

Example"Problem:"
!

Given"an"image,"search"through"all"possible"subEwindows"and" iden5fy"those"which"contain"a"human"face"

Computer'Vision''Fall'2011'

Face)Detection)Problem)

Computer'Vision''Fall'2011'

Computer'Vision''Fall'2011'

Why)is)pattern)recognition)hard?)
!

To"write"an" algorithm,"you" may"need"to" precisely"dene"a" human"face" Your"descrip5on" should"be" invariant"to"pose," 3D"rota5on," occlusion,"race," gender"" The"descrip5on" should"hold"for"all" faces"and"nothing" which"is"not"a"face" should"match"the" descrip5on"

Computer'Vision''Fall'2011'

IMPOSSIBLE?"

If"it"were"not"for"the"human"existence"proof,"we"would"have"given"up"a"long"5me"ago"

State)of)the)Art)in)Face)Detection)

Viola/Jones Face Detector (2001): Using implementation in OpenCV


Computer'Vision''Fall'2011'

Results of Schneiderman/Kanade Face Detector


Computer'Vision''Fall'2011'

Face)Detection)Demo)

Computer'Vision''Fall'2011'

Anda mungkin juga menyukai