COMP 508-Computer Vision

Fall 2017

Computer Engineering Department, Koç University

 Dr. Yucel Yemez

 

Office: Eng139   Phone: (0212) 338 1585

E-mail: yyemez@ku.edu.tr

Lectures: Tue & Th 14:30-15:45, ENG-B18

Office Hours: Tue & Th 15:00-16:00

 

 

Syllabus

Assignments

Projects

Course flow

Links

 


 

Course Description

COMP 508 is a graduate course to introduce the fundamentals of computer vision theory and practice. With recent developments in computing, transmission and display technologies,  2D/3D visual data have become commonplace in scientific, industrial and commercial arenas. The digital visual data are mostly reflections from real world and contain useful information. The main goal of computer vision is to analyze sensed images for extracting this information, to construct scene descriptions and knowledge representations, to recognize objects and thereby to make useful decisions about physical objects and scenes.

The course is open to graduate and (highly motivated) undergraduate students who are willing to understand the vision technology in conjunction with real world applications and especially very well suited to those who are interested in doing research in computer vision. Good programming skills and knowledge of Matlab/C/C++ are necessary for the course project and homework assignments. Basic DSP knowledge will also be very helpful.

Syllabus (in pdf)

Textbooks: No required textbooks, but reading from the following books will be helpful. 

Honor Code

All code and documentation handed in exams, assignments and projects must be your own work. In programming assignments, you can exchange ideas, but you should not ever share your code, even partly.

Grading

Final grades will be composed of: (tentative)

Programming assignments

30%

Exam

35%

Project

35%


TA

Emel Kupcu, Office: ENG-Z20, ekupcu@ku.edu.tr, Office hours: Tue-Wed 16:00-17:00


 

Programming assignments

There will be programming assignments posted here. I plan the first assignment to be in OpenCV/C. The other assignments will probably be in Matlab.


Assignment 1, due to 13 Oct, Friday.

Assignment 2, due to 30 Oct, Monday.
Assignment 3, due to 20 Nov, Monday.


 

Projects

By the first month of the semester, each student will have chosen a topic for her/his project. Projects can be either research oriented or applications programming oriented, addressing one of the computer vision problems/concepts/applications covered throughout the course. Depending on the chosen topic, students may be expected to do a literature survey on different techniques aiming at solving the specified problem and then to implement and test one of these techniques. A software implementation is mandatory, using OpenCV or Matlab.

Students are expected to submit a project proposal by Nov 10.

 


 

Lecture Notes

 

 

Lecture Topic Reading (from suggested textbooks)
21 Sep
Intro
Szeliski: Chapter 1
26 Sep
Imaging
Szeliski: Chapter 2, Sonka (3rd ed.): Ch. 1, 2, 3
28 Sep
Filtering
Szeliski: Ch. 3.1, 3.2  Sonka: Ch. 5.3
3-5 Oct
Frequency
Szeliski Ch. 3.4, Sonka Ch. 3
10-17 Oct
Feature Detection
Szeliski Ch. 4, Sonka Ch. 5.3
19 Oct
RANSAC
Szeliski Ch. 6.1.4, Ch. 9.1
26, 31 Oct, 2,7 Nov
Recognition
Szeliski Ch. 14, Sonka Ch 8,9,10
9-14 Nov
PCA_LDA
Szeliski Ch. 14.2.1
16, 21, 23 Nov
Segmentation
Szeliski Ch. 5,  Sonka Ch. 6
28 Nov - 14 Dec
3D Vision
Szeliski Ch 2.1, 6.3, 7, 11, 12       Sonka Ch 11,12
Also: Multiple View Geometry in Computer Vision, Hartley and Zisserman



 

Some Useful Links

OpenCV Resources