Stat 238

Vision as Bayesian Inference

Mon/Wed: 12:00-1:20 Winter 2012, Boelter 4413.


Course Description

This course models vision as Bayesian Inference. It concentrates on visual tasks such as segmenting images, detecting objects in images, and recognizing objects. Its goal is to describe the state of the art techniques. The handouts consist of copies of the lecture notes and related papers.

Reading Material

Grading Plan

3 homework assignments (25% each). Term project or review (25%).

Homework1
Homework2

Tentative Schedule

Lecture Date Topics Handouts Supplements
Additional Readings
1 01-09 Introduction
Lecture1
Supplement1

2
01-11 Images and Statistical Edge Detection
Lecture2
Notes2


01-16
Martin Luther King Holiday



3
01-18
Factorized Probability Models. Inavariant Features
Lecture3

KonishiYuille
4
01-23
Weak Membran Models
Lecture4
IntroProbGraphs
TV-norm
5
01-25
MRFs for Labeling: Gibbs and Mean Field Theory
Lecture5
MRFchapter
Grab-Cut
6
01-30
Exponential Models, ML learning, and MRFs
Lecture6
DellaPietra
ZhuWuMumford
7
02-01
Spectral Clustering, Super-pixels, and Edge detection
Lecture7
Superpixels

8
02-06
Unsupervised learning, and Dynamic Programming
Lecture8
DP&A*

9
02-08
Hidden Markov Models
Lecture9
HMMvision

10
02-13
Lambertian Lighting Models  Lecture10


11
02-15
Structure from Multiple Views
Lecture11
ManhattanWorld
SzeliskiDraft
12
02-20
Stereo and Belief Propagation Lecture12
Stereo
See MRF Chapter for BP
13
02-22
Segmentation and Image Parsing Lecture13

RegionCompetition
14
02-16
AdaBoost/Regression: Face and Text detection
Lecture14
ViolaJones
ChenYuille
15
02-21
Deformable Template Models of Objects
Lecture15

TuYuille
16
02-23
Learning Deformable Template Models Lecture16

YuilleHe
17
02-28
Active Appearance Models and Grammars
Lecture17
AAMs
FORMs
18
03-02
 Hierarchical Compositional Models  ReviewPaper
ActiveBasis


03-07





03-09