Computer Vision Number: CSD 15-385/685 Instructor: Srinivasa Narasimhan Teaching Assistants: Gunhee Kim Supreeth Achar Office Hours: Srinivasa: By Appointment Gunhee: Wed. 6pm-8pm Supreeth: Mon. 6pm-8pm Time: TR 10.30 am - 11.50 am Location: WeH 5409 (Lecture) NSH 3001 (Office Hours) Bulletin Board: cmu.cs.class.cs385 (Instructions here.)

Summary

This course provides a comprehensive introduction to computer vision. The areas that the course will cover are image processing; the physics of image formation; the geometry of computer vision (stereo reconstruction and tracking); and statistical methods for detection and classification.

Overview

Prerequisites This course requires familiarity with calculus, basic probability theory and linear algebra. MATLAB, the language of choice for the programming assigments will be covered as part of the introduction to the course. Successful completion of the following courses is required: 15-213/18-243 Introduction to Computer Systems and either 18-202 Mathematical Foundations of Electrical Engineering or both 21-241 Matrix Algebra, and 21-259 Calculus in Three Dimensions Once you've completed 15-385, you may be interested in other courses offered by the Carnegie Mellon Graphics Lab. Textbook The required textbooks for 15-385 this semester is: Horn, Berthold K.P. Robot Vision. The MIT Press, 1986. Assignments & Grading The grades for this course depend on 5 homework assignments and 1 midterm exam. In addition graduate students enrolled in the course will have to complete a final project. The weighting of marks is given below (homework and midterm exam weights will be scaled down to 80% for graduate students with the other 20% for the final project). (15%) Homework 1 (15%) Homework 2 (20%) Homework 3 (10%) Midterm (20%) Homework 4 (20%) Homework 5 (20%) Final Project (Graduate Students Only) You will be given a total of three "late days" for the semester. You may use these late days to extend the deadline of any programming or written assignment without penalty. Once all three late days have been used, further extensions will only be granted at the instructor's discretion and may incur a grading penalty. Note: Please use the Bulletin Board as a primary resource rather than emailing TAs directly. It will yield a more timely respones as all the TAs will be browsing it and your question may even have been answered already!

Syllabus

Note: This syllabus may change during the course. Keep checking back.   Part 0: Introduction Name: Introduction Date: Tue 01/17 Slides: Lecture 1 Slides Name: Working With MATLAB® Date: Thurs 01/19 Slides: Tutorial (thanks to David Kriegman and Serge Belongie) More tutorials thanks to Martial Hebert: Operations, Programming, Working with images   Part 1: Image Processing Name: 1D Signal Processing Date: Tue 01/24 Slides: Lecture 3 Slides Assignment: Assignment 1 Begins Assignment 1 Handout Name: Image Processing Date: Thurs 01/26 Slides: Lecture 4 Slides Name: Image Pyramids Date: Tue 01/31 Slides: Lecture 5 Slides Name: Edge Detection Date: Thurs 02/02 Slides: Lecture 6 Slides Name: Hough Transform Date: Tue 02/07 Slides: Lecture 7 slides Part 2: Physics Based Vision Name: Appearance and BRDF Date: Thurs 02/09 Slides: Lecture 8 Slides Name: Photometric Stereo Date: Tue 02/14 Slides: Lecture 9 Slides Assignment: Assignment 1 Due Assignment 2 Begins Assigment 2 Handout Name: Shape from Shading Date: Thurs 02/16 Slides: Lecture 10 Slides Reading: Ramachandran's paper Name: Direct and Indirect Illumination Date: Tue 02/21 Reading: CAVE Lab Project Website Reading: Paper   Part 3: Geometry - Tracking and Reconstruction Name: Image Formation Geometry and Projection Date: Thurs 02/23 Slides: Lecture 12 Slides Name: Optical Flow Date: Tue 02/28 Slides: Lecture 13 Slides Name: Image Alignment and Tracking Date: Thur 03/01 Slides: Lecture 14 Slides (Thanks to Iain Matthews for the slides) Reading: Lucas-Kanade 20 Years On Assignment: Assignment 2 Due (3/1 Thursday) Assignment 3 Starts Assignment 3 Handout Name: Midterm Review Date: Tue 03/06 Name: Midterm Exam Date: Thurs 03/08 Name: Binocular Stereo 1 Date: Tue 03/20 Slides: Lecture 15 Slides Name: Binocular Stereo 2 Date: Thur 03/22 Slides: Lecture 16 Slides Assignment: Assignment 3 Due Assignment 4 Starts Assignment 4 Handout Name: Structured Light Range Imaging Date: Tue 03/27 Slides: Lecture 17 Slides Name: Photo-Tourism and Internet Stereo Date: Thur 03/29 Weblink: Website with results and Google Tech Talk Part 4: Statistical Methods Name: Principal Component Analysis 1 Date: Tue 04/03 Slides: Lecture 19 Slides Name: Principal Component Analysis 2 Date: Thur 04/05 Slides: Lecture 20 Slides Name: Feature Detection (Blobs and SIFT) Date: Tue 04/10 Reading: Paper: Distinctive Image Features from Scale-Invariant Keypoints, David Lowe, IJCV 2004. Name: Feature Detection and Classification Date: Thur 04/12 Slides: Lecture 22 Slides Assignment : Assignment 4 Due Assignment 5 Starts Assignment 5 Handout Name: Classification Date: Tue 04/17   Part 5: Recent Research Name: Image Based Rendering Date: Tue 04/24 Slides: Lecture 22 Slides Notes: Reading Name: Novel Cameras Date: Thur 04/26 Slides: Lecture 23: Slides Name: Open Challenges in Computer Vision Date: Tue 05/01 Assignment : Assignment 5 Due   Finals and Projects Name: Graduate Student Project Presentations Date: Tue 05/08