CSE586/EE554 Computer Vision II
Mathematical Tools for Computer Vision
CSE Department, Penn State University
Instructor: Robert Collins

Spring 2015 Schedule: Tues/Thurs 9:45-11:00AM, IST 113

Course Syllabus

Textbooks: 1) Simon Prince, Computer Vision: Models, Learning and Inference, 2) Rick Szeliski, Computer Vision: Algorithms and Applications,

Presentation Order

Review of Probability and Parameter Estimation
Lecture Notes
Review of prob and statistics (Jan 13)   [notes]
Bayesian Parameter Estimation (Jan 15)   [notes]
Roadmap [notes] and MLE/MAP Bernoulli+Beta (Jan 20)   [derivation]
Reading Assignments
Read chapters 2-4 of the textbook
Homework 1 on MLE/MAP estimation of categorical distribution, and Dirichlet smoothing.
Dataset to use for part 3 of homework 1
Homework 1 due Sunday Feb 1. Upload as a pdf file in Angel dropbox. *10 percent off each day late*

Bag of Features Processing
Lecture Notes
Feature Extraction (Jan 22)   [notes] [6 per page]
Clustering: K-Means (Jan 27-29)   [notes] [6 per page]
K-means Derivation (Jan 27-29)   [notes]
Gaussian Mixture Models and EM (Feb 3)   [pdf] [6 per page]
Meanshift, Mediodshift, Quickshift (Feb 5)   [pdf] [6 per page]
Bag of Features Processing (Feb 10)   [pdf] [6 per page]
Reading Assignments
Read textbook chapters 13.1-13.3 and 7.4
Readings For Presentations and Critiques (Thurs Feb 12)
M.Hauser Video Google: A Text Retrieval Approach to Object Matching in Videos", Sivic and Zisserman 2003
K.Kim Visual Categorization with Bags of Keypoints", Csurka et.al. 2004
Y.Chen Object Categorization by Learned Universal Visual Dictionary, Winn et.al. 2005
T.Guo Discovering Object Categories in Image Collections, Sivic et.al. 2005
H.Zhong Sampling Strategies for Bag-of-Features Image Classification, Nowak et.al. 2006
Critique and Oral Presentation Instructions
Read all the papers. Prepare and submit written critiques for two of them. Here are some instructions on how to format your critique and tips for preparing an oral presentation. To give you an idea what I am looking for, please looks at this sample critique and sample oral presentation. Cavaet: the sample critique I've written would most likely receive a "check" in our check-minus, check, check-plus grading system. However, if you are trying for a check-plus, please don't go overboard... for example, keep the length to no more than 2 pages.
Critiques due in Angel Thurs Feb 12 BEFORE CLASS, NO LATE ACCEPTANCE! (because the whole point is to have read through and thought about the papers before seeing the presentations in class that day, so you can participate in the discussion.)
If you are giving a presentation, you don't have submit critiques for this set of papers.
Project 1: Bag of Visual Words
Implement and experiment with bag of visual words. We discussed the many tasks and design options involved in implementing bag of visual words at the end of the lecture on Feb 10. For convenience, that slide and more instructions about what to hand in are included here.

Regression and Classification
Lecture Notes
Intro: Learning and Vision (Feb 14,19)   [pdf]
Regression (Feb 19)   [derivation]   [slides]   [6 per page]
Viola Jones Detector / Boosting (Feb 24)   [slides]   [6 per page]
Dalal&Triggs HoG Detector / SVMs (Feb 26)   [slides]   [6 per page]
Deformable Part Models (Mar 3)   [slides]   [6 per page]
Reading Assignments
Read textbook chapters 8.1-8.4
Readings For Presentations and Critiques (Thurs, Mar 5)
XI, HAO Interactive Object Counting, Arteta et.al. 2014
YUE, TAIWEI Face Alignment at 3000 FPS via Regressing Local Binary Features, Ren et.al. 2014
DAVIS, KAMERAN Detecting Pedestrians Using Patterns of Motion and Appearance, Viola et.al. 2005
MORGAN, JOHN From Categories to Individuals in Real Time — A Unified Boosting Approach, Hall and Perona, 2014
LI, YUELONG Word Channel Based Multiscale Pedestrian Detection Without Image Resizing and Using Only One Classifier Costea and Nedevschi, 2014

Graphical Models
Lecture Notes
Graphical Models and Message Passing (Mar 17,19)   [slides]   [6 per page]
Hidden Markov Model and Kalman Filter (Mar 24)   [slides]   [6 per page]
Reading Assignments
Read textbook chapters 10 and 11
Readings For Presentations and Critiques (Thurs, Mar 26)
WU, MINGJIE Recognizing Human Activities from Sensors using HMMs Constructed by Feature Selection Techniques, Cilla et.al. 2009
WANG, JIA Hidden Markov Models for Face Recognition, Nefian and Hayes, and Face recognition using Hidden Markov Models, Samaria. (Consider these two papers as being one, for presentation and critiques.)
KARE, NACHIKET Pictorial Structures Revisited: People Detection and Articulated Pose Estimation, Andriluka et.al. 2009
ADVANI, S. Simple but Effective Tree Structures for DP-based Stereo Matching, Bleyer and Gelautz 2008
SU, WEI-KAI Probabilistic Temporal Head Pose Estimation Using a Hierarchical Graphical Model, Demirkus et.al. 2014
Readings For Presentations and Critiques (Thurs, April 2)
DAWARE, RAHUL An Application of Markov Random Fields to Range Sensing, Diebel and Thrun.
AMARO-RIVERA, Y. Efficient Mean Shift Belief Propagation for Vision Tracking, Park et.al. 2008
YANG, XIAO GrabCut - Interactive Foreground Extraction using Iterated Graph Cuts, Rother et.al. 2004
WOLFF, MARK Parsing World's Skylines using Shape-Constrained MRFs, Tonge et.al. 2014
SEYED MOUSAVI 3-D Depth Reconstruction from a Single Still Image, Saxena et.al. 2007

Sampling-based Methods
Lecture Notes
Intro to Sampling Methods (Apr 7)   [slides] [6 per page]
No class on Thurs Apr 9. Watch a good intro to MCMC while I'm gone.
Markov Chain Monte Carlo (Apr 14)   [slides] [6 per page]
Decision Trees and Ferns (Apr 16)   [slides] [6 per page]
Readings For Presentations and Critiques (Tuesday, Apr 21)
Critiques due also Tuesday Apr 21 in Angel, before class
REGE, ALISHA Tracking Multiple Humans in Crowded Environment, Tao and Nevatia
XU, RENMEI Finding People by Sampling , Ioffe and Forsyth
YIN, CHONGJIA Monte Carlo Localization for Mobile Robots, Dellaert et.al.
SHAH, HARSHIL Sampling Plausible Solutions to Multi-body Constraint Problems , Chenney and Forsyth. [see also sample animation results here.]
SAFI, JARIULLAH Visual Tracking by Sampling Tree-Structured Graphical Models , Hong and Han, ECCV 2014.

Shape Matching
Lecture Notes
Shape Alignment (LS, ICP, RANSAC, Hough) (Apr 23)   [slides] [6 per page]
Active Shape Models (Apr 28)   [slides] [6 per page]
Readings For Presentations and Critiques (Thursday, Apr 30)
Critiques due also Thurs Apr 30 in Angel, before class
LI, PEIXUAN Combined Object Categorization and Segmentation with an Implicit Shape Model, Leibe et.al.
VU, TIEP Object Detection using a Max-Margin Hough Transform , Maji and Malik
SHI, YONGJING Nonparametric estimation of multiple structures with outliers, Zhang and Kosecka
ASMUTH, MARK Robust Multiple Structures Estimation with J-linkage , Toldo and Fusiello.
WANG, SHUTING Shape Matching and Object Recognition using Low Distortion Correspondences , Berg et.al.