15-385/685 Computer Vision

SYLLABUS

[ Home | Syllabus | Links ]

Schedule:

Date  Lecture Topic  Reading  Assignments 
T 1/11 1. Introduction and Overview  
Image Processing  
R 1/13 2. Image filtering    
T 1/18 3. Feature Extraction     HW 1 out
R 1/20 4. Multi-scale analysis    
Object recognition  
T 1/25 5. Pattern classification    
R 1/27 6. Pattern description    
T 2/1   7. Bayes decision     HW 1 due, HW 2 out
R 2/3   8. Boosted cascade    
Physics of vision  
T 2/8 9. Imaging basics    
R 2/10 10. Imaging basics II      
T 2/15 11. Stereo and depth    
R 2/17 12. Shape from shading     HW 2 due, HW 3 out
Contour and Shape models  
T 2/22 13. Edges and contour    
R 2/24 14. Principal component analysis    
T 3/1   15. Shape models    
R 3/3   16. Mid-term examination   HW 3 due , HW 4 out.
M 3/7   Midterm grade due   . .
R 3/8,10   Spring Break   . .
Segmentation and Grouping  
T 3/15 17. Blind source separation    
R 3/17 18. Markov random field     .
T 3/22 19. Graph-cut     .
R 3/24 20. Hough and EM     HW4 due HW 5 out
Matching and Tracking  
T 3/29 21. Deformable matching      
T 3/31 22. Optical flow      
R 4/5 23. Tracking    
Biological visual system  
R 4/7 24. Early visual modules   HW 5 due
T 4/12   25. Cortical hierarchy  
R 4/14   No class Spring Carnival    
T 4/19   26. Contextual processing   FCE  
R 4/21 27. Active vision FCE  
Student Presentations  
T 4/26 28. Project Presentation  
R 4/28 29. Project Presentation    
S 4/30 30. Project Presentation    
M. 5/2: Project due (paper, codes)    
5/10 31. Final Examination   8:30-11:30 a.m.    
R 5/12 Final Grade due. 

Course Description

An intensive introduction to the theory and practice of computer vision, i.e. the analysis of the patterns in visual images with the view to understanding the objects and processes in the world that generate them. Major topics include feature extraction, image representation, grouping, discrimination, learning, inference, tracking and active vision. The emphasis is on the learning of mathematical concepts and techniques, and the application of these techniques to solve real vision problem. Students will learn to think mathematically, and develop skills in translating mathematical thoughts into computer programs. The discussion will be guided by comparision with human and animal vision, from psychological and biological perspectives.
 

Course Information:

Class location and time: Wean 5403. Tuesday, Thursday 3:00 p.m - 4:20 p.m.
Recitation: (optional) Matlab tutorials: place and time TBA.
Website: course info: http://www.cs.cmu.edu/afs/andrew/scs/cs/15-385/www
Course directory: homework submission: /afs/andrew.cmu.edu/scs/cs/15-385/students05/your_directory
Course Blackboard: Lecture nots/Handouts/solutions: https://blackboard.andrew.cmu.edu/
Required Readings: Lecture notes and handouts. No required textbook.
Reference Textbooks (on reserve): Forsyth, D. & Ponce, J. Computer Vision: a modern approach, Prentic Hall, 2002 (F)
Jain, R, Kasturi, R. & Schunck, B.G. Machine Vision , Wiley, 1995. (M)
Marr, D. Vision, Freeman, 1982. (M)
Palmer, S.E. Vision Science: Photons to Phenomenology. MIT Press. Cambridge, MA. 1999 (P)
Horn, B. Robot Vision, McGraw Hill, 1986 (H)
Duda, R., Hart, P.E., & Stork, D.G. Pattern Classification , 2001. (D)
Gonzalez, R. and Woods, R. Digital Image Processing, Addison-Wesley,1993.

Grading:

Assignments % of Grade (15385) % of Grade (15685) Topic
HW 1 12.5 10 Matching
HW 2 12.5 10 Recognition
HW 3 12.5 10 Inference
HW 4 12.5 10 Modeling
HW 5 12.5 10 Segmentation
Term Project 30 30 Proposal/Paper/Presentation
Examinations 20 20 Midterm/Final

Grading scheme:
88-100: A
75-87.9: B
60-74.9: C
0-60: F

15-385 vs 15-685:

15-385 is designed primarily for advanced undergraduate students. Graduate students and undergraduates (with permission) can take 15-685 for (12 units) graduate credit. 15-385 students are required to do only 4 of the 5 homework assignments, or the 4 highest grades will be counted. The term project for the graduate students is also expected to be more substantial.

You will lose 1 point for each class you miss after the midterm. No exception will be granted.

Homework:

Reports should be type-written if possible, but can be hand-written, except for printouts of Matlab code and program output. Templates (for doc and latex) will be provided. Students can collaborate with one partner on assignment 1-3. No collaboration is allowed on assignment 4 nd 6. Everyone should write a report with partner's name specified.

Projects:

For term project, a type-written report (hard and soft in pdf or ps format) is necessary. Collaboration with one classmate is allowed. Team of three requires special permission from the instructor. Matlab programs with comments should also be turned in to the course directory of one of the partner. The project write-up should be no less than 6 and no more than 10 pages in CVPR format, with additional pages allowed for commented Matlab codes as appendix. Project presentations will be arranged according to descending order of seniority. Master first, then seniors, and then juniors.

Late Policy:

Homework may be turned in late within one week, with a 10 percent discount of the grade. No extension is possible for term project.