Introduction to Computer Vision and Inference (ppt) - Nyu

unclesamnorweiganAI and Robotics

Oct 18, 2013 (3 years and 7 months ago)

80 views

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
1

Introduction

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
2

Vision


``to know what is where, by looking.’’
(Marr).


Where


What


© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
3

Why is Vision Interesting?


Psychology


~ 35% of cerebral cortex is for vision.


Vision is how we experience the world.


Engineering


Want machines to interact with world.


Digital images are everywhere.

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
4

Vision is inferential: Light

(http://www
-
bcs.mit.edu/people/adelson/checkershadow_illusion.html)

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
5

Vision is Inferential: Prior
Knowledge

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
6

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
7

Computer

Vision


Inference


Computation


Building machines that see


Modeling biological perception


© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
8

A Quick Tour of Computer
Vision

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
9

Boundary Detection

http://www.robots.ox.ac.uk/~vdg/dynamics.html

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
10

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
11

Boundary Detection

Finding the Corpus Callosum

(G. Hamarneh, T. McInerney, D. Terzopoulos)

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
12

Tracking

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
13

Tracking

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
14

Tracking

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
15

Tracking

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
16

Tracking

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
17

Stereo

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
18

Stereo

http://www.magiceye.com/

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
19

Motion

http://www.ai.mit.edu/courses/6.801/lect/lect01_darrell.pdf

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
20

Motion
-

Application

(www.realviz.com)

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
21

Pose Determination

Visually guided surgery

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
22

Recognition
-

Shading

Lighting affects appearance

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
23

Classification

(Funkhauser, Min, Kazhdan, Chen, Halderman, Dobkin, Jacobs)

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
24

Vision depends on:


Geometry


Physics


The nature of objects in the world


(This is the hardest part).

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
25

Approaches to Vision

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
26

Modeling + Algorithms


Build a simple model of the world


(eg., flat, uniform intensity).


Find provably good algorithms.


Experiment on real world.


Update model.

Problem:
Too often models are simplistic or
intractable.

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
27

Bayesian inference


Bayes law: P(A|B) = P(B|A)*P(A)/P(B).


P(world|image) =


P(image|world)*P(world)/P(image)


P(image|world) is computer graphics


Geometry of projection.


Physics of light and reflection.


P(world) means modeling objects in world.


Leads to statistical/learning approaches.

Problem:
Too often probabilities can’t be known and are
invented.


© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
28

Engineering


Focus on definite tasks with clear
requirements.


Try ideas based on theory and get
experience about what works.


Try to build reusable modules.

Problem:
Solutions that work under specific
conditions may not generalize.

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
30

The State of Computer Vision


Science


Study of intelligence seems to be hard.


Some interesting fundamental theory
about specific problems.


Limited insight into how these interact.

© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
31

The State of Computer Vision


Technology


Interesting applications: inspection,
graphics, security, internet….


Some successful companies. Largest
~100
-
200 million in revenues. Many in
-
house applications.


Future: growth in digital images exciting.


© 2004 by Davi Geiger

Computer Vision

January 2004 L1.
32

Related Fields


Graphics. “Vision is inverse graphics”.


Visual perception.


Neuroscience.


AI


Learning


Math: eg., geometry, stochastic processes.


Optimization.