EE5731 Visual Computing

Electrical and Computer Engineering, NUS





Description:

The goal of computer vision is to make computers work like human visual perception, namely, to understand and recognize the world through visual information, such as, images or videos. Human visual perception, after millions of years of evolution, is extremely good in understanding and recognizing objects or scenes. To have similar abilities to human visual perception (or beyond), computer vision scientists have been developing algorithms by relying on various visual information, and this course is about some of these algorithms. In case you are wondering why we should care about computer vision, consider this: if you think your visual perception system is important and beneficial, so is computer vision.


Textbooks: "Computer Vision: Models, Learning, and Inference", by S.J.D. Prince | "Multiple View Geometry", by R. Hartley and A. Zisserman | "Computer Vision: Modern Approach", by Forsyth and Ponce
Instructor: Robby T. Tan


Teaching Schedule:

The following schedule will not be strictly followed.


Date
Topic
Lecture Note
August 16 1. INTRODUCTION + IMAGE FORMATION

Reading: Additional Resources:
Lecture note 1

August 22 2. FACE DETECTION

Reading:
Lecture note 2
August 29 3. IMAGE FEATURES: HOG

Reading:
  • "Computer Vision" textbook Chapter 13: Image Preprocessing and Feature Extraction, particularly Sec. 13.1 (per-pixel transformations), and Sec. 13.3.3 (Histogram of Oriented Gradients)
Additional Resources:
Lecture note 3
September 5 4. IMAGE FEATURES: SIFT

Reading: Additional Resources:
Lecture note 4
September 12 5. IMAGE STITCHING

Reading:
  • "Computer Vision" textbook: Chapter 14: Sec. 14.3 (Homogeneous Coordinates),
  • "Computer Vision" textbook: Chapter 15, Sec. 15.1 (2D transformation models), Sec. 15.5 (transformation between images), Sec. 15.6 (robust learning of transformations), Sec. 15.7 (applications)
  • RANSAC: wikipedia
Additional resources:
  • Homography: wikipedia
  • Automatic Panoramic Image Stitching Using Invariant Features: pdf
  • Image Alignment and Stitching: A Tutorial: pdf

Lecture note 5
Assignment 1
September 19 6. CAMERA GEOMETRY

Reading:
  • "Multiple View Geometry" textbook: Chapter 6 (Camera Models): Sect.6.1 (finite cameras), Chapter 7 (Computation of the Camera Matrix): Sect. 7.1 (basic equations)
  • "Multiple View Geometry" textbook: Chapter 9 (Epipolar Geometry and Fundamental Matrix): Sect. 9.1 (Epipolar Geometry), Sect. 9.2 (The Fundamental Matrix)
Additional resources:
  • "Computer Vision" textbook chapter 14 (Pinhole Camera): sect. 14.1 (the pinhole camera), sect. 14.4 (learning extrinsic parameters), sect. 14.5 (learning intrinsic parameters)


Lecture note 6
September 26 RECESS WEEK

October 3 CLASS CANCELLED (Possible make-up class will be discussed later)



October 10 7. STEREO AND MARKOV RANDOM FIELD

Reading:
  • "Computer Vision" textbook: Chapter 12 (Models for grids): Sect. 12.1 (Markov Random Fields), Sect. 12.2 (MAP inference for binary pairwise MRFs)


Lecture note 7
October 17 8. DEPTH FROM STEREO

Reading:
  • Comments on Depth Map from a Video Sequence, UU-TR'11: [pdf]


Lecture note 8
Assignment 2 (CA2)
October 24 9. DEPTH FROM VIDEO

Reading:
  • Depth Map from a Video Sequence, TPAMI'09: [pdf]

See lecture note 8
October 31 10. OPTICAL FLOW (Part 1/2)

Reading:
  • Lucas-Kanade optical flow: wikipedia
  • Horn-Schunck optical flow: pdf
Additional reading:
  • Lucas-Kanade method: pdf

Lecture note 10
November 7 11. OPTICAL FLOW (Part 2/2)

See lecture note 10
November 14 12. TEXTURE-STRUCTURE DECOMPOSITION + REVIEW

Reading:
Lecture 12
FINAL EXAM





Syllabus: