CS 6476 Computer Vision
Georgia Tech
I. General Information (See Tabs for more Information)
Course Number: CS 6476-O01 (Spring 2024)
Course Name: Computer Vision
This course provides an introduction to computer vision including fundamentals of image formation, camera imaging
geometry, feature detection, and matching, stereo, motion estimation and tracking, image classification, and scene
understanding. We’ll develop basic methods for applications that include finding known models in images, depth
Description: recovery from stereo, camera calibration, image stabilization, automated alignment, tracking, and recognition.
The focus of the course is to develop the intuitions and mathematics of the methods in lecture, and then to learn about
the difference between theory and practice in the problem sets.
Program: Georgia Tech's Online MS in Computer Science
II. Team/People
Instructor Irfan Essa
Contact via Ed Discussion. Email for Private and Urgent Issues ONLY, please use CS6476 in subject line.
Video Lectures by Aaron Bobick
Do not contact Professor Bobick at all about the logistics of the class. You may contact him to thank him though.
Head IA Bob Kerner
Karan Samel
Head TAs
Nikolai Warner
TAs Abhijit Suprem
Anisha Pal
Eshani Chauk Contact via Ed Discussion. Email for
Shiva Gantha Private and Urgent Issues ONLY, Use Post
to Instructors OPTION in Ed Discussion.
Haris Hussain
Apoorva Beedu
Shreya Punjabi
Jack Kolb
Sakshi Nihatkar
Instructional
Arpan Chakraborty
Designers
Video Production Megan Smith
III. Important Websites
Grading, Schedule, Syllabus, and Final
Canvas https://gatech.instructure.com/courses/334646
Exam.
Google Doc (Syllabus This document PUBLISHED!
https://rb.gy/i96z6
/ Info / Schedule)
Exclusively used for discussions and
Ed (for Discussions) https://edstem.org/us/courses/43607/discussion/
contacting Instructors!
Notes:
1 All students are required to participate, attend to above websites. No EXCEPTIONS
No information will be shared via any other sites (G+, FB, etc.) or services (slack, twitter, reddit, etc.). Students are
welcome to create their own social media sites, but none of the instructors are required to be on those sites and will not
2
participate there regularly. Do note, that students should not share any of thw work product (assignments, projects, etc)
to a public site.
IV. Assignments & Grading
A. Assignments Type 1: There will be 1 assignments of this type 5.0%
B. Assignments Type 2: There will be 5 assignments of this type 65.0%
C. Project: Topics to choose from will be given near the release date 15.0%
D. Exam: Cumulative, timed, and online 15.0%
Total 100.0%
V. Books / Readings
Szeliski (2010), "Computer Vision: Algorithms and Applications", Springer,
1 SZ http://szeliski.org/Book/
2010
Forsyth & Ponce (2011), "Computer Vision: A modern approach", 2nd Ed.,
2 FP Publishers Site
Pearson 2011
Other readings maybe added
VI. Policies
A. Communications
WITH the Professor and TAs should be exclusively through Ed Discussion. No emails! Professor and TAs will do their best
1
to respond to questions within 2 days of posted question.
Ed Discussion will serve as the primary and ONLY source of communication and sharing announcements with the
2
students.
All communications should be professional and courteous. TA/Graders and Students are all required to maintain high
3
standards of interaction on Ed Discussion.
4 The online forum (Ed Discussion) is for course-related discussion. Not a forum to publically raise issues about the class. If
you have some issues, please raise them PRIVATELY via Ed Discussion just with the INSTRUCTORS!
B. Assignments
We will be using the class autograder for submitting the homework. Dates and Deadlines are counted by the final
1
submission timestamp.
Homework Assignments will be graded both with an autograder portion and a TA-graded portion, with a list of criteria
2 (specified on the assignment) such as quality of work, completeness, insight into technical issues, insight into other
relevant issues, etc.
Each assignment will be fully graded and returned USUALLY within two weeks of submission. Please allow two weeks to
3
pass to ask about the current grading status. If there is delay for some reason, it will be announced.
Over the course of the semester, you'll have 6 “free” late days to submit assignments PS1 through PS6. Late days do not
apply to the Late Submission Policy Quiz, the Plagiarism Policy Quiz, the Final Project or the Final Exam; none of these
assignments will be accepted late for any reason. Additionally, PS1 through PS6 will not be accepted more than 2 weeks
past their respective due date. Our intention is to give you some flexibility around your work commitments, family
obligations, vacations, and the like. Additional rules:
- For every extra late day (past the 6 “free” late days) used, you'll incur a penalty of 1% from your final grade. For
instance, if at the end of the course you used 8 late days in total, and your final grade was 91%, your score will be 89%
and Grade B after the penalty is applied. (In this example the penalty is 2% because 8 late days in total were used - 6 late
4 days = 2)
- Up to the first 6 days, your late submissions will be considered in hours (so you have 144 hours, that can be used over 6
assignments). However, a penalty is applied per day. ie. If your total late time is 6 days 1 hour or 6 days 23 hours, you'll
still incur a 1% penalty.
- Gradescope displays by how many hours your submission is late.
- If you submit the report late by an hour, and the code by 2, you would have used up to 3 hours of your late submission
quota.
If you have a medical or family emergency, please contact the Dean of Students who may grant an exception to the late
policy if your circumstances warrant it. We must receive approval from the Dean to grant an exception.
5 See collaboration policy below for more details on how to collaborate
Instruction included with the assignment and in the code provided MUST be explicitly followed, especially any and all
6
directions like how to submit and the file naming conventions specified
Regrade requests can be made via gradescope. Please provide clear details as to why you are requesting a regrade. All
7 regrade requests must be made within ONE (1) week of the grade release. For grades released in the last week of the
term, the regrade request must be made by the last day of the final exams week.
8 There will be no peer feedback this semester. We tried it, it did not work, so we are not using it.
9 All DUE dates will be on Canvas, and the timezone will be Anywhere on Earth Time (AoE) time. Please plan accordingly.
As we have a 6 assignments, there may be overlap on assignments. We expect students to manage their schedule to
10
meet the deadlines for each of the assignments
Students are welcome to work and submit assignments before their due date. The lectures will all be available from week
11 1. TAs will try to answer questions related to the assignments as much as they can, but most conversations maybe most
active as per the schedule planned for the class
Students are welcome to work and submit assignments before their due date. The lectures will all be available from week
12 1. TAs will try to answer questions related to the assignments as much as they can, but most conversations maybe most
active as per the schedule planned for the class
If the assignment does not follow the specific requirements, like using the REQUIRED template, the asssignemnt will be
13
returned UNGRADED with a score of ZERO
C. Discussions (via Ed All class discussions will be on the Ed Discussion site listed above. Here are some very specific guidelines for these
Discussion) discussions, which MUST be adhered to:
1 All posts must be professional and cordial and about/related to the course material at hand.
2 Students WILL not post specific answers to any of the assignments to Ed Discussion before the due date of said
assignments. In some instances, TAs will start a special discussion for students to share and discuss their assignments
after the DUE date.
3 Before asking a question on the Forum, students should search for an answer to their question. It most probably has
been discussed already
4 Instructor team will start weekly discussion threads about relevant topics. Before posting a new thread, please see these
threads and these official threads will be actively monitored by the Instructor team
5 Instructor team will attempt to answer all questions, as possible. But, please do NOT expect answers within hours. TAs
are instructed let students answer each others questions too, as that support more interactive learning.
6 Students can post annonymously to the class, but their IDENTITIES will be known by the instructor team
7 Instructor team is required to maintian privacy of all students, so please ensure that you communicate with them
privately (using the private channels via Ed Discussion) to communicate with them.
8 If there is a complaint about the class, please DO not post a public note to Ed Discussion. Please communicate directly
with the instructor team. We will do our best to address it. If it is NOT addressed, please use OMS Assistance (Point G.
below).
D. Websites Following are the websites we will OFFICIALLY use for this class:
1 Canvas: Grading and Final Exam.
2 Ed Discussion: For Official Announcements, and Forums for discussion.
3 Gradescope: Assignment Submission.
4 Google Docs: (This site) for syllabus/schedule and general information.
No information will be shared via any other site (G+, FB, etc.). Students are welcome to create their own social media
5
sites, but none of the instructors are required to be on those sites and will not participate there regularly.
E. Grading Grading Scale (for each assignment/unit and for the entire class).
1 Above 90%
2 80%-89.99%
3 70%-79.99%
4 60%-69.99%
5 Below 60%
F. Honor Code
1 All assigned work is expected to be individual, except where explicitly written
otherwise. You are encouraged to discuss the assignments with your classmates;
however, what you hand in should be your own work. If any work product was GT Honor Code
produced based on discussions with someone else (in the class OR outside), please
specify clearly in the final turn-in.
G. Collaboration
Policy
1 As stated above with the Honor Code, but worth making explicit here. Collaboration between students on work assigned
in class is fine. You are encouraged to discuss your work with each other. But each individual students MUST submit their
own work, done solely by themselves. In some cases, you may have had a fellow student or a non-student friend, help
you with an assignment or work (say to take a picture!). You are REQUIRED to acknowledge any help you may have
received in completing the work assigned, even as small as holding the light, or suggesting a possible path to a solution.
Please be explicit and provide details. We will be checking for code plagiarism in our assessment, so please NO copying
code from the Web/Internet.
2 Any code snippets must be cited and limited to a maximum of 5 lines. We understand you may not be familiar with
some libraries and APIs presented in this class and you will likely look up usage examples for individual functions. You
may study these examples but the code used in your assignment must be your own.
As part of this course’s grading process, any suspicion of copying will be reported to the Office of Student Integrity for
further analysis.
3 All students must also ensure that they DO NOT make any of the code for problem sets publicly available and are
required to take steps to prevent future students from having access to it. Consequently, if you're using any version
control systems such as git, please make sure that you mark your repositories as private.
4 You may not collaborate at all on the final exam. Students are not to discuss any questions or answers from the actual
exam with classmates or anyone else until after the testing period is complete.
H. OMS Assistance If after contacting your TA and the instructor you do not feel your issue has been resolved, you may escalate the issue by
emailing oms-advising@cc.gatech.edu and asking that your ticket to be assigned to Jay Summet.
Lectures Assignments Readings
# Begin [1] Title Lesson Topic Dura Title Release Due Text
Module #
PS #
tion
(min)
[2]
DUE DATES provided here for just for planning purposes. ACTUAL due dates are in CANVAS.
Time Zone for this class is Atlanta TIME
1 8-Jan 1A Introduction 1A-L1 Introduction 42 1 Images as Functions 8-Jan 22-Jan SZ: 1.1, 1.2
2A Linear image processing 2A-L1 Images as functions 44
2A-L2 Filtering 24 FP: 4
2A-L3 Linearity and convolution 34
2A-L4 Filters as templates 13
2A-L5 Image to image projections 27
2A-L6 Edge detection: 2D operators 19
2 15-Jan 2B Hough Transforms 2B-L1 Hough transform: Lines 36 FP: 10.1
2B-L2 Hough transform: Circles 13
2B-L3 Generalized Hough transform 16 pdf
2C Frequency domain analysis 2C-L1 Fourier transform 36 FP: 4
2C-L2 Convolution in frequency 22
domain
2C-L3 Aliasing 35
3 22-Jan 3A Camera models 3A-L1 Cameras and images 33 2 Traffic Lights and Signs 19-Jan 5-Feb FP: 1, 2.1-2.2
3A-L2 Perspective imaging 26
3B Stereo geometry 3B-L1 Stereo geometry 26 FP: 7
3B-L2 Epipolar geometry 11
3B-L3 Stereo correspondence 29
4 29-Jan 3C Camera calibration 3C-L1 Extrinsic camera calibration 24 FP: 8
3C-L2 Instrinsic camera calibration 16
3C-L3 Calibrating cameras 31
5 5-Feb 3D Multiple views 3D-L1 10 3 Adventures in AR 19-Jan 19-Feb
3D-L2 Homographies and mosaics 33
3D-L3 Projective geometry 14
3D-L4 Essential matrix 22
3D-L5 Fundamental matrix 37 pdf
6 12-Feb 4A Feature detection 4A-L1 Introduction to "features" 13 FP: 5.3-5.4; SZ: 4
4A-L2 Finding corners 39
4A-L3 Scale invariance 23 pdf
4B Feature descriptors 4B-L1 SIFT descriptor 27 FP: 5.4; SZ: 4.1
4B-L2 Matching feature points (a little) 16
4C Feature robustness 4C-L1 Robust error functions 31 494
4C-L2 RANSAC 33 FP: 10.2-10.4
7 19-Feb 5A Photometry 5A-L1 Photometry 35 4 Motion Detection 16-Feb 4-Mar
5B Lightness 5B-L1 Lightness 26
5C Shape from shading 5C-L1 Shape from shading 34
8 26-Feb 6A Motion 6A-L1 Introduction to motion 16 FP: 9, 10.6
6B Optical flow 6B-L1 Dense flow: Brightness 24
constraint
6B-L2 Dense flow: Lucas and Kanade 17
6B-L3 Hierarchical LK 33
6B-L4 Motion models 24 FP: 11.3
9 4-Mar 7A Tracking 7A-L1 Introduction to tracking 14 Object Tracking and
5 1-Mar 18-Mar
Pedestrian Detection
7B Parametric models 7B-L1 Tracking as inference 21
7B-L2 The Kalman filter 36
10 11-Mar 7C Non-parametric models 7C-L1 Bayes filters 23
7C-L2 Particle filters 17 FP: 11.5
7C-L3 Particle filters for localization 24
Lectures Assignments Readings
10 11-Mar
# Begin [1] 7C Non-parametric
Titlemodels Lesson Topic Dura Title Release Due Text
Module #
PS #
tion
(min)
[2]
7C-L4 Particle filters for real 15
11 18-Mar 7D Tracking considerations 7D-L1 Tracking considerations 27 6 Classification 15-Mar 1-Apr
8A Recognition / Classification 8A-L1 Introduction to recognition 21 Final Project Topic FP: 16
15-Mar
release
8B Classification: Generative 8B-L1 Classification: Generative 28 FP: 15.1-15.2
models models
8B-L2 Principle Component Analysis 48 FP: 16.1.5
8B-L3 Appearance-based tracking 26
12 25-Mar 8C Classification: Discriminative 8C-L1 Classification: Discriminative 27
models models
8C-L2 Boosting and face detection 27
8C-L3 Support Vector Machines 51
8C-L4 Bag of visual words 14 FP: 20.1
13 1-Apr 8D Action recognition 8D-L1 Introduction to video analysis 24 7 Final Project 22-Mar 24-Apr
8D-L2 Activity recognition 32
8D-L3 Hidden Markov Models 46
14 8-Apr 9A Color spaces and 9A-L1 Color spaces 36
segmentation
9A-L2 Segmentation 18
9A-L3 Mean shift segmentation 18
9A-L4 Segmentation by graph 13
partitioning
15 15-Apr 9B Binary morphology 9B-L1 Binary morphology 37
9C 3D perception 9C-L1 3D perception 34
10A The retina 10A-L1 The retina 38
10B Vision in the brain 10B-L1 Vision in the brain 27
16 22-Apr Last Week FINAL EXAM 24-Apr 1-May
[1] Monday
[2] (min.)
