This course will cover
essentials of Computer Vision, a discipline
that strives to develop techniques to help computers "see" and
understand images. The course is of interest to anyone
seeking to process images and acquire a general background in problems
related to real-world perception, object and scene
recognition and 3D reconstruction. The geometric aspects of the course will focus on extracting
3D metric information from 2D images. The second theme
covers methods for extraction of semantic information. This
will entail problems of image classification, object detection
(e.g. how to detect people, cars or other object of interest in
images), activity recognition. Aplications to 3D modelling, video analysis,
video surveillance, image based retrieval, object detection and recognition and
vision based control will be discussed.
Grading Homeworks (about every 2 weeks) 50% Exam:
30% Final project: 20%
Prerequisites linear algebra, calculus, CS 580 or CS 584 and CS 583
Lecture Materials Lecture slides, lecture notes provided by instructor
Recommended Textbooks, Resources
[1] Invitation to 3D Vision: From Images to Geometric Models: Y. Ma, S. Soatto, J. Kosecka and S. Sastry web site
[2] Computer Vision: Algorithms and Applications. R. Szeliski, 2010, Springer online version of the book
[3] Computer Vision: A Modern Approach: D. Forsythe and J. Ponce, Prentice-Hall, 2003
[4] Image Processing, Analysis, and Machine Vision. Sonka, Hlavac, and Boyle. Thomson.
[5] Computer Vision. Ballard and Brown web site
Required Software
Python, OpenCV
Course Outcomes
Basic knowledge of image formation process
Basic knowledge of image processing techniques for color and gray level images: edge detection, corner detection, segmentation
Basics of video processing, motion computation and 3D vision and geometry
Basics of image classification, object detection and recognition video processing
Ability to implement basic vision algorithms in Python/OpenCV (open source computer vision library)
Ability to implement image classification and object detection with convolutional neural networks using Pytorch library
Ability to apply the appropriate technique to a problem, write a project report and present the results in class.
Academic Integrity:
The integrity of the University community is affected by the individual choices made by each of us. GMU has an Honor Code with clear guidelines regarding academic integrity. Three fundamental and rather simple principles to follow at all times are that: (1) all work submitted be your own; (2) when using the work or ideas of others, including fellow students, give full credit through accurate citations; and (3) if you are uncertain about the ground rules on a particular assignment, ask for clarification. No grade is important enough to justify academic misconduct. Plagiarism means using the exact words, opinions, or factual information from another person without giving the person credit. Writers give credit through accepted documentation styles, such as parenthetical citation, footnotes, or endnotes. Paraphrased material must also be cited, using MLA or APA format. A simple listing of books or articles is not sufficient. Plagiarism is the equivalent of intellectual robbery and cannot be tolerated in the academic setting. If you have any doubts about what constitutes plagiarism, please see me.
CS department Honor Code can be found here.