1f55c5c06d
7.9 KiB
CSE 559A: Computer Vision
Course Overview
This course introduces computational systems that analyze images and infer physical structure, objects, and scenes. Topics include color, shape, geometry, motion estimation, classification, segmentation, detection, restoration, enhancement, and synthesis. Emphasis is on mathematical foundations, geometric reasoning, and deep-learning approaches.
Department: Computer Science & Engineering (559A) Credits: 3 Time: Tuesday/Thursday 1–2:20pm Location: Jubel Hall 120 Modality: In-person
Instructor: Prof. Nathan Jacobs Email: jacobsn@wustl.edu (Piazza preferred) Office: McKelvey Hall 3032 or Zoom Office Hours: By appointment
TAs: Nia Hodges, Dijkstra Liu, Alex Wollam, David Wang Office hours posted on Canvas.
Textbook
Primary: Computer Vision: Algorithms and Applications (2nd Ed.) — http://szeliski.org/Book/ Secondary: Computer Vision: Models, Learning, and Inference — http://www.computervisionmodels.com/ Secondary: Foundations of Computer Vision (MIT Press)
Prerequisites
Official: CSE 417T, ESE 417, CSE 514A, or CSE 517A Practical: Python programming, data structures, and strong background in linear algebra, vector calculus, and probability.
Learning Outcomes
Students completing the course will be able to:
- Describe the image formation process mathematically.
- Compare classical and modern approaches to geometry, motion, detection, and semantic tasks.
- Derive algorithms for vision problems using mathematical tools.
- Implement geometric and semantic inference systems in Python.
Course Topics
- Low-Level Feature Extraction: classical and modern (CNNs/Transformers)
- Semantic Vision: classification, segmentation, detection
- Geometric Vision: image formation, transformations, motion, metrology, stereo, depth
- Extended Topics: e.g., generative models, multimodal learning (TBD)
Grading
Homework
Homework consists of ~7 programming assignments in Python, focused on implementing core algorithms. Most include auto-graded components. Two late days allowed per assignment; after that, the score is zero. No late submissions for quizzes, paper reviews, or project.
Exams / Quizzes
There are ~5 quizzes covering lectures and readings. They include both theoretical and applied questions. No late quizzes.
Paper Reviews
Four short reviews of recent computer vision research papers. Includes an in-class discussion component.
Project
An individual or small-team project implementing, evaluating, or developing a vision method. Specifications on Canvas.
Final Grades
| Component | Weight |
|---|---|
| Homework (~7) | ~60% |
| Quizzes (~5) | ~15% |
| Paper Reviews (4) | ~5% |
| Project | ~15% |
| Participation | ~5% |
Grading Scale
| Letter | Range |
|---|---|
| A | 94% and above |
| A- | <94% to 90% |
| B+ | <90% to 87% |
| B | <87% to 84% |
| B- | <84% to 80% |
| C+ | <80% to 77% |
| C | <77% to 74% |
| C- | <74% to 70% |
| D+ | <70% to 67% |
| D | <67% to 64% |
| D- | <64% to 61% |
| F | <61% |
Schedule
Approximate; see Canvas for updates.
| Week | Date | Topic | Notes |
|---|---|---|---|
| W1 | Jan 14 | Overview | |
| Jan 16 | Image Formation & Filtering | ||
| W2 | Jan 21 | Image Formation & Filtering | HW0 due |
| Jan 23 | Image Formation & Filtering | ||
| W3 | Jan 28 | Image Formation & Filtering | HW1 due |
| Jan 30 | Image Formation & Filtering | Module Quiz | |
| W4 | Feb 4 | Deep Learning for Image Classification | Paper review due |
| Feb 6 | Deep Learning for Image Classification | ||
| W5 | Feb 11 | Deep Learning for Image Classification | HW2 due |
| Feb 13 | Deep Learning for Image Classification | ||
| W6 | Feb 18 | Deep Learning for Image Classification | HW3 due; paper review due |
| Feb 20 | Deep Learning for Image Classification | ||
| W7 | Feb 25 | Deep Learning for Image Classification | Module Quiz |
| Feb 27 | Deep Learning for Image Understanding | ||
| W8 | Mar 4 | Deep Learning for Image Understanding | HW4 due; paper review due |
| Mar 6 | Deep Learning for Image Understanding | Module Quiz | |
| Spring Break | — | — | |
| W9 | Mar 18 | Feature Detection, Matching, Motion | |
| Mar 20 | Feature Detection, Matching, Motion | ||
| W10 | Mar 25 | Feature Detection, Matching, Motion | HW5 due; project launch; paper review due |
| Mar 27 | Feature Detection, Matching, Motion | Module Quiz | |
| W11 | Apr 1 | Multiple Views and Stereo | HW6 due |
| Apr 3 | Multiple Views and Stereo | ||
| W12 | Apr 8 | Multiple Views and Stereo | |
| Apr 10 | Multiple Views and Stereo | Module Quiz | |
| W13 | Apr 15 | Extended Topic | HW7 due |
| Apr 17 | Extended Topic | ||
| W14 | Apr 22 | Extended Topic | Final project due this week |
| Apr 24 | Final Lecture / Presentations | No Final Exam |
Technology Requirements
Assignments may require GPU access (e.g., Google Colab, Academic Jupyter). Students must avoid modifying starter code in ways that break auto-grading.
Collaboration & Materials Policy
Discussion is allowed, but all submitted work (code, written content, reports) must be individual. Posting assignment solutions publicly is prohibited. Automated plagiarism detection (e.g., Turnitin) may be used.
Generative AI Policy
- Homework & Projects: Allowed with limitations; students must understand all algorithms used.
- Quizzes: May be used for explanation but not direct answering.
- Reports: AI-assisted writing permitted with responsibility for correctness.
More detailed rules are on Canvas.
University Policies
Recording Policy
Classroom activities and materials may not be recorded or distributed without explicit authorization.
COVID-19 Guidelines
Students with symptoms must contact Student Health for testing. Masking policies may change depending on conditions.
If you'd like, I can also produce:
- a cleaned, typographically polished version
- a PDF or GitHub-ready README.md
- a version styled identically to your department’s standard syllabus format