Computer Vision - Spring 2008
Instructor: Rene Vidal E-mail: rvidal@cis.jhu.edu

Mondays and Wednesdays, 1:30-2:45 p.m. 316 Hodson

Office Hours: Fridays 3-4 p.m. 302B Clark Hall

TA: Rizwan Chaudry, rizwanch@cis.jhu.edu

TA Office Hours: Tuesdays, 2:00 - 3:00 p.m. 317 Clark Hall

Course Description
This course gives an overview of fundamental methods in computer vision from a computational perspective. Methods include computation of 3-D geometric constraints from binocular stereo, motion, texture, shape-from-shading, and photometric stereo. Edge detection and color perception are studied as well. Elements of machine vision and biological vision are also included.

 

Announcements

  1. 05/09/2008: TA Office hours to collect/check final exams (All in Clark 317):
    1. Friday, 05/09/2008: 2 p.m. to 5 p.m.
    2. Monday, 05/12/2008: 2 p.m. to 4 p.m.
  2. 05/09/2008: A solution to the final can be found HERE. Please have a look at it before coming to contest your marks for the final.
  3. 05/05/2008: Review session for the final on Tuesday 05/06/2008 in Shaffer 202 at 4 p.m.
  4. 05/02/2008: Homework 4 solution is up!
  5. 05/01/2008: Homework 3 solution is up!
  6. 04/23/2008: You can find all the individual frames of the project video in PNG format HERE
  7. 04/22/2008: A few important announcements regarding the project presentations:
    1. Every student in every group has to present for 5 minutes each. Hence there would be 10 minutes per group of two students and 15 minutes for the group of three students. After the presentation, 5 minutes will be left for questions for every group. Students are advised to STRICTLY adhere to the 5 minute limit per student otherwise marks will be deducted for every minute over the limit.
    2. The presentations will be held in class Monday April 28 and Wednesday April 30. Every group is required to be ready to present their work on Monday. The groups that will eventually present on Monday will be chosen by a draw in class. This makes it fair for all the students to have the same amount of time to prepare the presentations.
    3. A member from every group is advised to come half an hour before class on both days with their laptops so that their laptops can be tested to work with the presentation equipment in the class rooms. If your laptops do not work at the time of your presentation, any time that you take to reconfigure/trouble-shoot/reboot etc will be taken out of your presentation time.
    4. The detailed 6 page report (title, abstract, intro, problem description, proposed solution, experimental evaluation, conclusions, references), would be due Wednesday April 30, 6 p.m. in class. Your report must conform to the same document style as in this document. i.e. it must be 6 pages, double column, single line spacing and Times New Roman font size 10 document. These specifications MUST be adhered to and failure will result in loss of marks. Microsoft Word and LaTeX templates can be found HERE
  8. 04/16/2008: Newly added to Useful Computer Vision resources at the bottom of the page:
    1. Dictionary of Computer Vision and Image Processing
    2. HIPR-2: Image Processing Learning Resources
  9. 04/10/2008: The project description can be found under Project in the Grading section.
  10. 04/08/2008: Here is the final list of the groups for the final project.
    Group Students
    1 Gagan Bansal, Sandeep Mullur
    2 Christopher McFarland, Tiffany Chen
    3 Giancarlo Troni, Raphael Sznitman
    4 Bogdan Vigaru, Michael Kutzer
    5 Samantha Mercer, Aru Sahni, Chuan Huang
    6 Steve Swedish, Steve Barolak
    7 Pin Wu, Sneha Verma
    8 Ehsan Elhamifar, Roberto Tron
    9 Jeffrey Cheng, Kapil Dalwani
    10 Yue Xing, Yin Chen
    11 Eric Lin, Chris Ecker
  11. 04/02/2008: Check the Handouts section for a very good reference for Lucas Kanade based Tracking.
  12. 04/01/2008: IMPORTANT: The full Indian traffic videos for the Project can be found HERE (compressed) and HERE (uncompressed - WARNING 500 MB file)
  13. 04/01/2008: An updated solution of Homework 1 with complete figures is available now. Please check the updated link.
  14. 03/26/2008: For the final project, if you have already decided the person you are going to work with (at most 2 persons per group), then email the TA your name and your group partner's name. For people who have not found a group partner yet, here's a current list of people looking for group partners. Once you have decided who you want to work with, email the TA your names so that your name can be removed from the list of people looking for partners to the list of decided groups. If you do not see your groups then email the TA ASAP.
  15. 03/26/2008: A solution to Homework 1 can be found HERE. Some diagrams for Q 2(a) are missing and will be updated ASAP.
  16. 03/24/2008: A solution to homework 2 can be found HERE.
  17. 03/24/2008: Class timings and locations have changed and there are several changes. Please check the individual class timings and locations in the table.
  18. 03/14/2008: Assignment 4 has been posted!
  19. 03/08/2008: For people who could not finish their filter implementation in Assignment 2, here is the code for the S, LM and RFS filters and a function that gives the output of the filters when applied to an image. It is not necessary that you use this code if your own code works fine.
  20. 03/05/2008: Assignment 3 has been updated again. Please download the latest version.
  21. 03/04/2008: Assignment 3 has been updated. Please download the new version.
  22. 03/02/2008: Assignment 3 has been posted!
  23. 02/18/2008: The updated Assignment 2 is now AVAILABLE.
  24. 02/17/2008: Assignment 2 is temporarily unavailable and will be updated soon.
  25. 02/16/2008: Assignment 2 has been posted!
  26. 02/06/2008: Assignment 1 has been updated and question 1(a) has been replaced. Please download the updated version of the assignment from the Homeworks section below.
  27. 02/01/2008: Assignment 1 has been posted!

 

Class notes

Most of the slideds used in class are based on slides available on the web at the following websites:

  1. Prof. Gregory Hager: Computer Vision, Johns Hopkins University, Fall 2006
  2. Prof. Ko Nishino: Introduction to Computer Vision, Drexel University, Winter 2008
  3. Profs. David Forsyth and Jean Ponce - Computer Vision: A Modern Approach book slides
  4. Prof. Pietro Perona, Visual Recognition, Caltech, Spring 2007
  5. Prof. Steve Seitz, Computer Vision, Washington University, Winter 2008
The copyright of the slides belongs to the respective authors.
Lecture No. Date Time Venue Class notes Handouts Slides
1 01/28/2008 1:30 p.m. - 3:45 p.m. 316 Hodson SVD SVD and Least Squares Handout Syllabus
2 01/30/2008 1:30 p.m. - 3:45 p.m. 316 Hodson Least squares, Optics and Homogeneous coordinates   Introduction
3 02/04/2008 1:30 p.m. - 3:45 p.m. 316 Hodson Optics and Photometry   Cameras
4 02/06/2008 1:30 p.m. - 3:45 p.m. 316 Hodson   Photometry  
5 02/11/2008 1:30 p.m. - 3:45 p.m. 316 Hodson     Color
6 02/13/2008 1:30 p.m. - 3:45 p.m. 316 Hodson     Filtering
7 02/18/2008 1:30 p.m. - 3:45 p.m. 316 Hodson Signal processing   Filtering
8 02/20/2008 1:30 p.m. - 3:45 p.m. 316 Hodson Edge Detection   Edge Detection
9 02/25/2008 1:30 p.m. - 3:45 p.m. 316 Hodson Edge Detection and Textures   Texture
10 02/27/2008 1:30 p.m. - 3:45 p.m. 316 Hodson Segmentation and Clustering   Textons
11 03/03/2008 1:30 p.m. - 3:45 p.m. 316 Hodson More on KMEANS   Segmentation
12 03/05/2008 1:30 p.m. - 3:45 p.m. 316 Hodson   Spectral Clustering Segmentation
13 03/10/2008 1:30 p.m. - 3:45 p.m. 316 Hodson   Spectral Clustering Segmentation
14 03/12/2008 1:30 p.m. - 3:45 p.m. 316 Hodson     Segmentation
  Spring Break          
15 03/24/2008 6:00 p.m. - 7:15 p.m. 315 Hodson Motion   Optical Flow
16 03/26/2008 6:00 p.m. - 7:15 p.m. 110 Clark Optical Flow Lucas Kanade Tracking Matching
17 03/31/2008 6:00 p.m. - 7:15 p.m. 315 Hodson Homography   Features and RANSAC
18 04/02/2008 6:00 p.m. - 7:15 p.m. 110 Clark   SIFT Tutorial, David Lowe's SIFT Paper Features and RANSAC
19 04/07/2008 6:00 p.m. - 7:15 p.m. 315 Hodson Homography    
20 04/09/2008 6:00 p.m. - 7:15 p.m. 110 Clark Structure from Motion    
21 04/14/2008 6:00 p.m. - 7:15 p.m. 315 Hodson Structure from Motion    
22 04/16/2008 6:00 p.m. - 7:15 p.m. 110 Clark  
  1. Recognition for Eigenfaces: The original CVPR paper by Turk and Pentland
  2. A nice walkthrough of the Eigenfaces procedure
Recognition (slides 1)
23 04/21/2008 6:00 p.m. - 7:15 p.m. 315 Hodson     Recognition (slides 2)
24 04/23/2008 6:00 p.m. - 7:15 p.m. 110 Clark  
  1. The Representation and Matching of Pictorial Structures
  2. Pictorial Structures for Object Recognition
  3. Coarse-to-Fine Face Detection
 Recognition (slides 3)
25 04/28/2008 6:00 p.m. - 7:15 p.m. 315 Hodson     Project
26 04/30/2008 6:00 p.m. - 7:15 p.m. 110 Clark     Project

 

Textbooks

  1. Forsyth and Ponce (FP): Computer Vision a Modern Approach
  2. Trucco and Verry (TV): Introductory Techniques for 3-D Computer Vision
  3. Shapiro and Stockman (SS): Computer Vision

 

Syllabus

  1. Cameras
    1. FP1, TV 2.2.2, SS 2.2
  2. Photometry
    1. Radiometry: FP 4, TV 2.2.3, SS 2.1
    2. Shading: FP 5, TV 9, SS 6.6
    3. Color: SS6, FP6
  3. Image Enhancement and Filtering
    1. Enhancement: SS 5.1-5.2
    2. Filtering: FP 7, 8.1-8.2, TV 3, SS 5.3-5.4
    3. Edge detection: SS 5.6-5.8, FP 8.3, TV 4.2
    4. Corner detection: TV 4.3
  4. Texture
    1. FP9, SS7, TV 9.5.1
  5. Image Segmentation
    1. K-means and EM: FP 14.4
    2. Spectral clustering and NCut: FP 14.5
    3. Color Segmentation: SS 6.5
    4. Texture Segmentation: SS 7.4
  6. Motion, Optical Flow, Image Registration and Matching
    1. TV 8
  7. Structure from Motion
    1. Camera Calibration: FP 2-3, TV 2.4, 6
    2. Stereo: FP 11
    3. Two-View Geometry: FP 10.1
    4. Affine SFM: FP12
  8. Object Recognition

 

Grading

  1. Homeworks (40%)
    1. HW1: Shape from Shading (DUE DATE: February 15)

      Face Data, Sphere Data

      SOLUTION

    2. HW2: Image Filtering, Edge Detection, and Texture Classification (DUE DATE: February 29)

      CURET dataset

      SOLUTION

    3. HW3: Image Segmentation (DUE DATE: March 14)

      4. Images for Intensity segmentation

      Images for Color segmentation

      Images for Texture segmentation

      Datasets for clustering

      SOLUTION

    4. HW4: Image Mosaicing (DUE DATE: April 4)

      5. Images for Lucas Kanade Tracker

      Russian glass plates

      Mosaicing problem data sets: Set 1 Set 2 (Right click on these links and choose "Save As" to save the MAT files)

      SOLUTION

  2. Project (30%)
    1. Description: title and problem description (April 9)
    2. Progress report: 3 pages (title, abstract, intro, problem description, proposed solution) (April 18)
    3. Presentations: 10 min + 3 min questions (April 28-30)
    4. Final report: 6 pages (progress report + proposed solution, experimental evaluation, conclusions, references) (April 30)
  3. Exam (30%)
    1. Thursday May 8th (2-5PM)

 

Administrative

  • Late policy:
    • Homeworks and projects are due on the specified dates.
    • No late homeworks or projects will be accepted.
  • Honor policy:

    The strength of the university depends on academic and personal integrity. In this course, you must be honest and truthful. Ethical violations include cheating on exams, plagiarism, reuse of assignments, improper use of the Internet and electronic devices, unauthorized collaboration, alteration of graded assignments, forgery and falsification, lying, facilitating academic dishonesty, and unfair competition.

  • Homeworks and exams are strictly individual
  • Projects can be done in teams of two students

 

Useful Computer Vision Resources