cv ch0 syllabus 2014

22
Computer Vision and Its Applications (in Home Video Surveillance) What to be covered? What is better to have as background? How will your grade be evaluated? Projects, exercises, exams, and others Jison Hsu Artificial Vision Laboratory Taiwan Tech

Upload: killua142

Post on 12-May-2017

217 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: CV Ch0 Syllabus 2014

Computer Vision and Its Applications (in Home Video Surveillance)

What to be covered?What is better to have as background?

How will your grade be evaluated?Projects, exercises, exams, and others

Jison HsuArtificial Vision Laboratory

Taiwan Tech

Page 2: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

About Me and My TA• Jison Hsu, E1-445, Ext. 3234

– My interests: face recognition, pedestrian detection and tracking, object recognition from images/videos;

– Office hours (OH): 13:30~15:00 on Tue.

[email protected]

• TA: – Hank Liu, E1-246, Ext. 3738, OH: 13:00~15:00 Fri.;

[email protected]

Page 3: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Textbooks and References• Main Textbook

– Computer Vision:Algorithms and Applications, by Richard Szeliski(2010, Sep.) (free e-book: http://szeliski.org/Book)

– OpenCV v2.4.8 documentation: http://opencv.itseez.com/

• Others– Pattern Classification, by R. O. Duda, P. E. Hart, D. G. Stork, Wiley-

Interscience; 2 ed. (2000)– Computer Vision: A Modern Approach, by D. Forsyth and J. Ponce,

Prentice Hall (2002);– Digital Image Processing, R. C. Gonzalez, R. E. Woods, Prentice Hall;

3 ed. (2007)– Very important: www.wikipedia.org, key in any topics or key words,

and enjoy the guided tour, for example, “computer vision” or “OpenCV”.

Page 4: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Lectures• Classification on image features

– Principal Component Analysis and Linear Discriminants,

– Bayesian Classifiers,

– SVM and Neural Networks,

– Adaboost Classifiers

– Clustering, if time allows!

Page 5: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

What Backgrounds Are Needed?

• It is better to know the basics of image processing, such

as color space, pixel, histogram, contrast, and we will

only give a brief review.

• You must know Matlab and C, otherwise this course

will be highly challenging to you.

Page 6: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

How You Are Graded?• Midterm and final exams: 13% each, tentative schedule:

4/17 and 5/29;

• 4 exercises: 6% each;

• Course project: 50%, oral proposal on 3/27 and final

presentation on 6/12.

• Incentive points are given to those who are responsive

in my lectures.

Page 7: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Course Project (1/3)• Topic: You can define your own project as long as it is

related to “Classification” or “Clustering”• Examples

– Face detection and recognition– Pedestrian and vehicle detection and tracking– License plate detection, segmentation and recognition– Object recognition– Image segmentation

Page 8: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

• I can offer you the database needed for your study if your choice is from those in boldface.

• Proposal presentation will be on 3/27, you must present your topic, the database to evaluate the performance, the information that you have collected and considered helpful to your project, such as codes and reference sites.

• If your project can be solved by some open source codes, you have to specify what you want to add in or compare it with other approaches.

• Your group must be ≤ 2 members.

Course Project (2/3)

Page 9: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

• To help you make progress, you will have to talk to me in my office hours and update me your status on your project. The schedule will be released on BB in the first week of March.

• To avoid repeated topics, those who have decided their topics and maybe methods can send a write-up to TA. We will post it on BB so that those who submit late must consider the differences from those who submit early. – The write-up must include topic and tentative method(s).

Course Project (3/3)

Page 10: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Lab Exercise• PC-based DVR with cable cameras (GeoVision)

– Functions and settings (F&S) including motion detection, (event triggered) recording, video playback;

Page 11: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Reports on Lab Exercises• Contents: (1) introduction to the system, (2) functions and

settings, (3) operations, (4) your executions and findings, (5) pros and cons, (6) suggestions to improve the system layout and functionality.

• No absolutely correct answers because this is NOT a question-and-answer type of assignment, but a study for you to get to know a system and learn something from it.

Page 12: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

• Survey Computer Vision related products– Example :

•WhatsMyHeartRate– http://facion.net/WhatsMyHeartRate/– Demo Video

•WorldCard– http://www.penpower.com.tw/product.asp?sn=109– Demo Video

Survey Exercises

Page 13: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

• Contents: (1) introduction to the system, (2) video demo (by yourself), (3)functions and settings, (4) operations, (5) your executions and findings, (6) pros and cons, (7) suggestions to improve the system layout and functionality.

Reports on Survey Exercises

Page 14: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

OpenCV Exercises• Familiarization of the tools in OpenCV

– Officially launched in 1999, the OpenCV project was originally an Intel Research initiative to advance CPU-intensive applications - http://en.wikipedia.org/wiki/Open_CV

• Download and install OpenCV v2.4.8– http://opencv.itseez.com/

• You are requested to complete 2 exercises using OpenCV. Samples using OpenCV will be provided. Each exercise will be released on the Blackboard (BB), and due in 2 weeks.

Page 15: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Reports on OpenCV Exercises• The first exercise is “people detection”• The second exercise is “PCA object or face recognition”

– Your report must include the following:• For people detection

– Input, output, command-line window or GUI. For example, a GUI with inputs of 50 still images and 2 videos (> 2mins) from the web, outputs are the cropped humans (for people detection).

• Flowcharts with an intro to functional calls or APIs from OpenCV;

•Results of your test with various settings on the parameters of the APIs and test images.

Page 16: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Upload Your Exercises• Upload your report to BB on 3/13, 4/10, 5/1, and 5/22

before class. It will be graded and available on BB two weeks later. Any delay will degrade your score accordingly.

Page 17: CV Ch0 Syllabus 2014

Computer Vision

Pattern Recognition Lab at E1-248

E1

2F

Lab opening hours: 13:00~17:30 on Tue, Wen, Thur, Fri.User manuals would be available on Blackboard. TA will be in E1-246.

You have to sign in and sign out each visit to the lab. A sign-in booklet is available in the lab.

Page 18: CV Ch0 Syllabus 2014

Computer Vision

Pattern Recognition Lab at E1-248

Booklet here

Page 19: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Introduction to Machine Vision

http://www.youtube.com/watch?v=TTnho9-i6dI&hd=1 http://www.youtube.com/watch?v=VjcJnCkoWUM&hd=1

Page 20: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Introduction to Computer Vision

http://www.youtube.com/watch?v=715uLCHt4jE&hd=1

Page 21: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

License Plate Recognition System

https://www.youtube.com/watch?v=El_AdmmTL24&hd=1

Page 22: CV Ch0 Syllabus 2014

Computer Vision, Js Hsu

Computer Imaging: Dan Connors at TEDxMileHigh

https://www.youtube.com/watch?v=ty5QtslaAHM