CENG 509

Vision Based Tracking and Modeling

This course covers the tracking of object and camera positions from images and videos by using computer vision techniques. Course contents include the mathematical theory and the algorithms used in practice necessary for modeling the objects and scene to be tracked.

Course Objectives

To provide the students with the necessary mathematical background required for basic object and camera tracking topics as well as information about the algorithms employed by state-of-the-art approaches.

Recommended or Required Reading

Kalman Filtering: Theory and Practice using MATLAB, 4th Edition, 2014 ,Multiple View Geometry by R. Hartley and A. Zissermen, 2nd Edition, 2003

Learning Outcomes

1. Describe the fundamental theoretical concepts in computer vision based tracking

2. Identify and discuss the state-of-the-art object tracking approaches

3. Design and implement a complete object tracking approach

4. Implement high performance real-time software for object tracking

Topics
Introduction to Tracking
Optical Flow and Lucas-Kanade Tracker
Template Tracking Basics
Advanced Template Tracking
Kalman Filter I
Kalman Filter II
Particle Filters
Introduction to Modeling
Sparse Models
3D Modeling
Semi-Dense models
Simultaneous Localization and Mapping I
Simultaneous Localization and Mapping II
Relocalization Approaches

Grading

Homework: %30

Research Presentation: %40

Final Exam: %30