|
|
Computer Vision CS
6320, Spring 2018 |
''Man Drawing a Lute'', woodcut by Albrecht Dürer, 1525 Metropolitan Museum of Art, New York |
|
|
|
|
Lectures: Instructor:
TA: Grading:
Prerequisites:
|
Monday & Wednesday, 1:25 – 2:40 PM, WEB L114 Srikumar Ramalingam, srikumar@cs.utah.edu Office hours: Monday and Wednesday 3:00
-4:00 PM MEB 3464 Programming assignments (60%), Project (20%), Final (20%) linear algebra, vector calculus, C/C++, Matlab, |
Class No
|
Date
|
Title
|
1
|
01/09
|
Introduction
to Computer Vision
|
2
|
01/11
|
Camera
Models and Image Formation
|
-
|
01/16
|
No Class!
|
|
|
01/17
|
HW1 (PDF, Latex) Released, Due on
01/29
|
3
|
01/18
|
Camera
Pose Estimation and RANSAC
|
|
|
01/23
|
3D
Reconstruction
|
5
|
01/25
|
Epipolar Geometry
|
6
|
01/30
|
Epipolar Geometry (continued)
|
7
|
02/01
|
The first meeting on projects!
HW2 (PDF,
Latex & Data) Released,
Due on 02/17
|
8
|
02/06
|
Keypoints and Descriptors
|
9
|
02/08
|
Image Matching (Slides, Paper)
|
10
|
02/13
|
An
Introduction to Graphical Models
|
11
|
02/15
|
Belief Propagation
|
|
|
02/19
|
Project Proposal
Due (Team Members, Problem Statement)
|
-
|
02/20
|
No Class!
|
12
|
02/22
|
Belief Propagation
(continued)
|
|
|
02/25
|
HW3 (PDF,
Latex), Due on 03/03
|
13
|
02/27
|
Belief
Propagation (continued)
|
14
|
03/01
|
Graph Cuts
|
|
|
03/05
|
Project Status Due (Introduction and Prior
Art)
|
15
|
03/06
|
Midterm Review
|
16
|
03/08
|
Midterm
|
-
|
03/13
|
No class! (Spring Break)
|
-
|
03/15
|
No class! (Spring Break)
|
17
|
03/20
|
Graph Cuts (continued)
HW4,
Due on 04/02
|
18
|
03/22
|
Using neural nets
to recognize handwritten digits (Slides, Chapter1)
|
19
|
03/27
|
Using neural nets
to recognize handwritten digits (continued)
|
20
|
03/29
|
Project Status
Report and Presentations
Intermediate report
due on 04/07
|
21
|
04/03
|
Project Status
Report and Presentations
|
22
|
04/05
|
How the backpropagation works (Slides, Chapter2)
|
|
|
04/07
|
HW5,
Due on 04/21
|
23
|
04/10
|
Improving the way the
neural networks learn (Slides,
Chapter3)
|
24
|
04/12
|
Improving the way the
neural networks learn (Slides,
Chapter3)
|
25
|
04/17
|
Visual proof that neural
nets can compute any function
Why are
deep neural networks hard to train
|
26
|
04/19
|
Stereo and
Semantic Segmentation |
27
|
04/24
|
Final Project Presentations (4+2 minutes for each group) |
28
|
04/26
|
|
29
|
05/01
|
Final Exam (6-8 PM)
|
|
|
05/05
|
Final Project Report Due
|
Project:
The project carries 25 points, and it should clearly demonstrate your understanding and interest in computer
vision. You are allowed to work in groups of
maximum size 2, but each one should clearly
state his/her contribution in the final report. Please use the latex template
from http://iccv2017.thecvf.com/files/iccv2017AuthorKit.zip.
·
Problem statement (5 points)
·
Algorithm and Experimental Results (15
points)
·
Report (5 points)
Homework:
There will be a total of 5
homework assignments. For getting
the full credit, please submit the solutions before the deadline. The assignments
will be graded for 90% of the points for
up to 2 days after the deadline.
You can discuss the general ideas and
concepts in homework problems freely, but you are expected to write the
solutions on your own. Please do not read or look at the answers or code of
other people. Presenting the work of someone else’s solution as your own will
be considered cheating. For a detailed description of the university’s policy
on cheating and academic integrity, please refer to http://regulations.utah.edu/academics/6-400.php.
Programming:
The project and the homework assignments
would require some programming, and we
encourage the students to use C++ and OpenCV
libraries. Matlab (See Matlab primer) can also be used.
Recommended
Reading:
There is no single textbook for this course.
Some of the lectures will use specific chapters from the following materials:
1.
CVBOOK: Computer
Vision: Algorithms and Applications by Rick Szeliski 2010 (available
online)
2.
DLBOOK1: Deep Learning (available online) by Ian Goodfellow,
Yoshua Bengio, and Aaron Courville, An MIT Press book, 2016
3.
DLBOOK2: Neural Networks and Deep Learning
(available online) by Michael Nielson
4.
CVNOTES1: Some lecture notes on geometric
computer vision (available online) by Peter Sturm
Additional
Reading:
Multiple View Geometry in Computer
Vision by Hartley and Zisserman 2004
The latest
research topics in computer vision can be
learned from the publications in the following computer vision
conferences:
· International Conference on Computer Vision (ICCV)
· Computer Vision and Pattern Recognition (CVPR)
· European Conference on Computer Vision (ECCV)