COMP 4301: Computer Vision

This course is an elective for both the Smart Systems and Visual Computing and Games Streams.

Lab In addition to classes, this course has six 3-hour structured laboratory sessions per semester.

Prerequisites:  COMP 3301 or Engineering 7854 or permission of the instructor

Availability: This course is usually offered once per year, in Fall or Winter.

Course Objectives

This course studies how to develop methods that enable a machine to “understand” or analyze images. The course introduces the fundamental problems in computer vision and the state-of-the- art approaches that address them. Topics include feature detection and matching, geometric and multi-view vision, structure from X, segmentation, object tracking and visual recognition.

Representative Workload
  • Assignments (2) 20%
  • In-class Exam 20%
  • Project 30%
  • Final Exam 30%
Representative Course Outline
  • Unit 1: Grouping and fitting (4 hours)
    • K-means
    • Hough transform
    • RANSAC
  • Unit 2: Feature detection and matching (4 hours)
    • Interest point detection (corners/blobs)
    • SIFT
    • HOG
  • Unit 3: Geometric and multi-view vision (4 hours)
    • Geometric transformation
    • Camera model and camera calibration
    • Image stitching
  • Unit 4: Feature based alignment (4 hours)
    • 2D and 3D feature based alignment
    • Pose estimation
  • Unit 5: Structure from X (4 hours)
    • Epipolar geometry
    • Stereo vision
    • Essential and fundamental matrix
    • Structure from motion
  • Unit 6: Segmentation and tracking (4 hours)
    • Foreground segmentation in video
    • Optical flow
    • Tracking
  • Unit 7: Recognition (4 hours)
    • Introduction to recognition
    • Object detect and recognition (face detection, pedestrian recognition)
    • General category recognition (bags of features)
Notes
  • Credit cannot be obtained for both Computer Science 4301 and Engineering 8814.

Page last updated May 24th 2021