CS 664 Computer Vision – Spring 2008
Cornell University

Class TR 2:55-4:10, 315 Upson

Professor: Dan Huttenlocher
4133 Upson
Office Hours, Wednesday 1-2pm
dph "at" cs.cornell.edu

Brief overview:

This course is intended for graduate students and advanced undergraduates who are interested in processing image and video data, in order to extract information about the scene that is being imaged. There is no textbook for the course. Handouts and papers will be made available online. A recommended text is Forsyth and Ponce’s book “Computer Vision: A Modern Approach”, but it covers topics that we won’t and vice versa.

The course has a more algorithmic flavor than many introductory computer vision courses. We will focus on efficient algorithms, precise problem definitions and methods that work well in practice.

We use material from various areas of algorithms and mathematics as well as requiring programming assignments, but this course does not teach algorithms, mathematics or programming. Thus we expect that students have good programming skills (using C or C++), a good mathematics background, and knowledge of algorithms. Students will be expected to pick up new mathematical and algorithmic techniques during the semester, as covered in lecture, and to relate the concepts from lecture to the programming assignments.

Assignments:

Assignment 1, filtering and edge detection
Assignment 2, project proposal

To hand in an assignment, login at http://cms.csuglab.cornell.edu with your Cornell netid and password and go to assignment 1 to upload a single zip or tar file with your source, executable and writeup.

Data for the final project:

Data files are available at http://web3.cs.cornell.edu/cs664/. The README file provides information about the data files and sensor configuration. For those who might care, the camera lens is Pentax Model#: C30405TH (4.8mm F/1.8).

Movies corresponding to the raw data files may also be useful, and are available as log1, log2part1, log2part2 and log2part3. You may need the FFDShow codec to view them.

Course outline:

Here is an outline of the topics to be covered and approximate schedule. This schedule (and possibly also the topics) will be updated during the semester. Readings will be handed out in class and added to the schedule as we go along.

1. Jan 22	Intro
2. Jan 24	Filtering	Edge Detection Handout, Wells paper
3. Jan 29	NO CLASS
4. Jan 31	Edge detection	Edge Detection Handout
5. Feb 5	Corner Detection (Crandall lecturer)
6. Feb 7	2D Geometry/Transforms
7. Feb 12	Interest points – SIFT features	Lowe SIFT paper
8. Feb 14	Distance Transforms
9. Feb 19	Matching: Chamfer and Hausdorff	Hausdorff Matching paper
10. Feb 21	3D camera geometry
11. Feb 26	3D camera geometry (cont’d)
12. Feb 28	Stereo	Scharstein and Szeliski paper
13. Mar 4	CLASS CANCELLED
14. Mar 6	Markov Random Field models
15. Mar 11	MRF Inference – Loopy BP	Felzenszwalb loopy bp paper
16. Mar 13	MRF Inference – Graph Cuts (Crandall)	Boykov et al paper
17. Mar 25	Visual motion/optical flow
18. Mar 27	Parametric motion
19. Apr 1	Structure from motion	Tomasi and Kanade paper
20. Apr 3	Some Recent Advances
21. Apr 8	Image Segmentation	Felzenszwalb segmentation paper
22. Apr 10	Image Segmentation	Weiss eigenvector segmentation paper
23. Apr 15	Face Recognition, Subspace Methods (Crandall)
24. Apr 17	Object Category Recognition, k-fans (Crandall)	Crandall k-fans paper
25. Apr 22	Pictorial Structures and fast algorithms
26. Apr 24	Image Matching and Robust Fitting
27. Apr 29	Recognition using latent SVM’s	Felzenszwalb et al paper
28. May 1	TBD

Course Requirements:

There will be a few short in-class quizzes, two assignments and a final project. The assignments and project will require programming, testing with image or video data, and a well thought-out write-up explaining what was done and what was learned.

The programming is probably best done in C or C++ due to the availability of libraries such as OpenCV, but Matlab can also be an option (Java is not recommended).