CAP 4453 Robot Vision Spring 2020 Home Page
Instructor: Dr. Niels da Vitoria Lobo
email: email@example.com (put CAP4453 and your name in subject line)
Office Location: HEC-252
Office Hours: Mon/Tue/Wed 11:45am -- 1pm
Office Phone: 407-823-2873 (407-UCF-CURE)
TA's Office Hours: ????day 1pm and ????day 2pm
If student can't make TA's hours, call the main TA (???-???-???3,
??????????????, email: ???????????????@knights.ucf.edu;
secondarily call ???;
s/he will reply, and make arrangements to meet at a mutually convenient time.
How to do the assignments
Update 22: Mar 25: Optical Flow Assignment (Required):
Here is the set of directions for installing OpenCV and doing the
Optical Flow demo.
In this assignment, you will run the Optical Flow algorithm from the
OpenCV Library. Follow the directions in the provided pdf notes (above) to
see how to install Visual Studio 2017, the OpenCV Library, and the
code that will call the appropriate modules from the library.
You are expected to show a live demo of your camera running the
optical flow (if this is impossible, you are allowed to run the
algorithm on two frames, as shown in the pdf directions). The Optical Flow
algorithm has been implemented in the library by following the theory that
is presented in the paper (bouguetopticalflow.pdf in Update 21).
These equations in the paper are similar to the ones discussed in
our Optical Flow notes (how to get results for 3 or more points), but
are derived in a different manner, and use a Pyramidal approach that is
also described in the paper. (You do not need to understand that paper to
obtain good experimental results).
For showing your working demo, you will do a live
demo with the grader, and will show that your cam is able to see a moving
pattern, such as some text on a sheet of paper (that is moved slowly),
and the cam window displays the optical flow vectors.
Update 21: Mar 24:
Update 20: Mar 19:
Now that this class has moved to an on-line mode, please also check the
Notifications that I send to Webcourses for the CAP4453 class.
So, regularly (every 8 hours or so), please check here for Updates, and
read the webcourses notifications. If you have not yet heard from me your Test
Score in a phone call, you must immediately send me email with a time that
I can call you or can communicate with you by webcourses Conference (which will
allow us to talk with audio; to set up the webcourses
Conference, you need to stay on-line after sending me an email that
you are ready to spend time focusing on
communicating with me). If you are stranded outside the USA, or do not have
a phone, we will be able to communicate adequately by webcourses Conference.
Update 19: Feb 15:
In addition to this list of questions, there will be a few single-point
questions that will test if the student has understood deeper issues and details
about the study material.
Update 18: Feb 12:
Practice Test Question Two and Project Preparation
Update 17: Feb 6 :
Additional Opportunity time to have Sobel.c and Canny.c graded:
at room HEC-234.
Friday Feb 7 , from 11:00am till 1pm. There will be more times later
in future weeks.
Update 16: Feb 5:
Convolutional Neural Net slides
Update 15: Feb 5:
Intro Neural Net slides
Update 14: Feb 3 :
Opportunity times to have Sobel.c and Canny.c graded: Both times are
The two times are: Wednesday Feb 5 , from 9:00am till 10:15am;
and Wednesday Feb 5, from 12:15pm till 2pm.
Update 13: Feb 3 :
Optical Flow slides 4 :
Update 12: Feb 3 :
Optical Flow slides 3 :
Update 11: Jan 29:
Optical Flow slides 2 :
Update 10: Jan 29:
Optical Flow slides 1 :
Update 9: Jan 22:
Required Assignment Two: In this assignment, you will download and run
our implementation of AdaBoost, using the positive and negative examples
we have provided.
Carefully read the steps in here,
Unzip the zip file.
Everything ( vboost, vdetect, positive and scenery)
is included in the .zip file (positive and scenery are in ..\AdaBoost\vboost,
Germany.pgm is in ..\AdaBoost).
Should you run into any issues while compiling the software, or during
running it, send an email to firstname.lastname@example.org.
For judging whether you are on the right track to success with this assignment,
run the test on the German Soccer Team image. After all the improvements
(such as training on 2000 images of each class),
your results should be such that most of the faces are found, and there are few
False Positives (boxes drawn in places where there are no faces). There is
no partial credit for this Assignment, you must succeed fully to
get the points. For grading, you will meet the grader and show the program
running and getting the result on an image that we will provide you.
Update 8: Jan 15:
For the face detection topic, Google: viola jones robust real-time object
Then, select the paper that is found at
www.cs.cmu.edu/~efros/courses/LBMV07/Papers named viola-IJCV-01.pdf
Update 7: Jan 15:
Update 6: Jan 13:
Update 5: Jan 8:
Update 4: Jan 8:
Update 3: Jan 6:
Due to Federal Government rules
related to financial aid, you must go to
Webcourses and complete Assignment Zero before the end of
this coming Friday, Jan 10, 2020.
Update 2: Jan 6 :
Update 1: Jan 6 :