高等電腦視覺 Advanced Computer Vision (922 U3910) - 2019 Spring

[Announcement] [Info] [Homeworks] [Class Materials] [Reference Materials]


Announcement

Homeworks Term Project Demonstration Term Project Proposal Final Exam Midterm Exam

Info

授課教授: 傅楸善教授
E-mail: fuh@csie.ntu.edu.tw
Web: http://www.csie.ntu.edu.tw/~fuh
Tel: (02)23625336 轉 327
上課時間: 週二 下午789節 (2:20PM ~ 5:20PM)
上課地點: 資訊工程系館(德田館) R105
辦公室時間: 週二 11:00AM ~ 12:00PM

TA
助教: 昝亭甄
E-mail: r06922029@ntu.edu.tw
TA時間: 週二 3:00PM ~ 5:00PM R328,可另寄E-mail詢問(主旨請加上 [ACV] )

Grading
Homework (3 times): 20%
Term Project: 30%
Midterm Exam: 20%
Final Exam: 30%

Textbook and References
Textbook [1] Richard Szeliski, Computer Vision: Algorithms and Applications, 2010.
References [2] Trevor Darrell's CS 280 Computer Vision class at Berkeley.
[3] Antonio Torralba's 6.869 Advances in Computer Vision class at MIT.
[4] Michael Black's CS 143Introduction to Computer Vision class at Brown.
[5] Kristen Grauman's CS 378 Computer Vision class at UT Austin.
[6] Alyosha Efros' 15-463 Computational Photography and 16-721 Learning-Based Methods in Vision classes at Carnegie Mellon.

Instructor of Each Chapter
CH01: 傅楸善
CH02: 曾柄元
CH03: 黃季軒
CH04: 董子嘉
CH05: 李建慶
CH06: 林炳彰
CH07: 楊博凱
CH08: 王嘉會
CH09: 張蓉蓉
CH10: 呂英弘
CH11: 黃泓叡
CH12: 許展銘
CH13: 郭冠軒
CH14: 呂彥穎


Homeworks

Regulations
  1. Homeworks are to be turned in ON TIME by FTP (site: 140.112.31.83, username: 2019cv2, password: 2019cv2, port: 12000) before class starts.
  2. Please compress the homework files to a rar or zip file. The filename format is rxxxxxxxx_hwx_vx.rar(ex. r05944036_hw1_v1.rar).
  3. Late homeworks will be subjected to grade penalties (20% off).
  4. Only electronic submissions are allowed. The homework should include two parts, the report and source code. The file format is restricted to Adobe PDF format.
    Failure to fulfill the file formats when submitting your homework will result with no grades. Also, please take note of the fields for submission.
  5. Your homework report should include the following contents: a description of your homework, the algorithm you used, your parameters (if any), your principal code fragment, and the resulting images (please paste the images in your report file).
  6. Any programming language are available for homework. You must not use any available libraries beyond image I/O (reading or writing image files from/to the disk/memory).
  7. Do not copy homeworks from others. Copying is cheating, and cheating is shameful. You and the person who let you copy his/her homework will both get a 0. In addition, you could be subjected to an on site demonstration for the following assignments.

Homework 1 Image Matching (Detecting Motion Vectors)
  • See handouts Chapter 1.
  • Images: trucka.im, truckb.im (bitmap version: trucka.bmp, truckb.bmp)
  • Detect motions vectors between trucka.im and truckb.im.
  • Use trucka.im as the basis, sample it by an 8×8, 11×11, 15×15, 21×21, 31×31 block.
  • Threshold of search range: 50 pixels. (This is a reference value only!)
  • Dimension of truck is 386×386 with 216 bytes of leading header.
  • Due date: 2019/03/05 (二) 2:20PM
Homework 2 Camera Calibration
  • See handouts Chapter 2.
  • Calculate the equivalent distance of the real world in 1 pixel of the digital camera.
  • You may have to find the parameters of the digital camera you are using.
  • Please specify the parameters you are using, i.e., focal lenght, object length, etc.
  • Please tabulate your result in your report, there may be error between the true "theoretical" value and the value you measured, but don't worry.
  • If you have no camera, you can borrow the camera and take photo at room 328, or using the sample at http://goo.gl/pWBNN0.
  • Due date: 2019/03/19 (二) 2:20PM
Homework 3 Calculating Optical Flow
  • See handouts Chapter 8.
  • Implement Horn & Schunck optical flow estimation.
  • Synthetically translate lena.im one pixel to the right and downward.
  • Try λ of 0.1, 1, 10.
  • Harvard vision format lena.im.
  • Windows bitmap vision format lena.bmp.
  • Due date: 2019/05/14 (二) 2:20PM
Term Project
Objective
  • A working prototype with new, original, novel ideas
  • Not just literature survey
  • Not just straightforward implementation of existing algorithms
  • All right to modify existing algorithms
Problems
  • Tentative term project problems: Exercises in textbook.
  • All right to be the same problem with Master's thesis.
Submit
  • Proposal: Submit one page in English explaining method, steps, expected results.
  • Due date: 2019/04/16 (二) 5:20PM
  • Report progress every other week.


Class Materials

Textbook
Handouts (PPT Format)

Research (pwd: 2019cv2)


Reference Materials