Course information
This course introduces the fundamental problems and methods of computer vision. We start with how images are formed and represented, and cover filtering, edge and feature detection, feature matching, image alignment and motion estimation. The second half of the course focuses on visual recognition with machine learning and deep learning: convolutional networks, object detection, segmentation, recurrent models, video and human action understanding, and recent vision transformer and vision-language models. Students will implement core algorithms in Python.
Instructor and time
Reference books
- Computer Vision: Algorithms and Applications, 2nd editionRichard Szeliski, 2022
- Foundations of Computer VisionAntonio Torralba, Phillip Isola and William T. Freeman, 2024
- Computer Vision: A Modern Approach, 2nd editionDavid Forsyth and Jean Ponce, 2011
- Dive into Deep LearningAston Zhang, Zachary C. Lipton, Mu Li and Alexander J. Smola
Grading
| Component | Weight |
|---|---|
| Project | 20% |
| Midterm | 35% |
| Final exam | 45% |
Announcements
All announcements and communication will be carried out via Piazza. The enrolment link and lecture notes will be shared there at the start of the semester.
Schedule
Tentative. Dates and lecture notes will be added during the semester.
| Week | Topic | Notes |
|---|---|---|
| 1 | Introduction and image formation | – |
| 2 | Image filtering and the frequency domain | – |
| 3 | Edge detection | – |
| 4 | Feature detection and description | – |
| 5 | Feature matching, image alignment and RANSAC | – |
| 6 | Motion estimation | – |
| 7 | Midterm | – |
| 8 | Machine learning for visual recognition | – |
| 9 | Convolutional neural networks | – |
| 10 | Object detection and segmentation | – |
| 11 | Recurrent models | – |
| 12 | Video and human action understanding | – |
| 13 | Vision transformers and vision-language models | – |
| 14 | Project presentations | – |