Syllabus

Course Description

This course is a project-based introduction to computer vision: how machines acquire, process, and interpret visual information from images and video. We begin with image formation, filtering, and local features, then build to the deep learning methods behind modern vision systems, including convolutional networks, vision transformers, and vision-language models. The course pairs conceptual understanding with hands-on experiments, so you learn not only how these methods work, but why they work and when they fail.

Every assignment asks for evidence, not only an implementation: you will run experiments, read the resulting curves, test methods on data you collect yourself, and defend your choices. The final project applies the same habits to a question of your own choosing.

Prerequisites (minimum grade of C in each course unless noted):

• I&C SCI 6D
• MATH 3A or I&C SCI 6N
• MATH 2B or AP Calculus BC (minimum score of 4)
• I&C SCI 46
• One of COMPSCI 112, COMPSCI 116, COMPSCI 171, or COMPSCI 178

In practice you’ll rely most on linear algebra, probability, and programming in Python.

Student Learning Outcomes

Upon successful completion of the course, students should be able to:

1. Explain the core concepts of computer vision, from image formation, filtering, and feature extraction to modern deep learning representations, and analyze how visual information is encoded and processed.
2. Implement, train, and evaluate computer vision models using modern frameworks such as PyTorch, and interpret their performance using appropriate metrics and diagnostics.
3. Critically assess the strengths, limitations, and failure modes of computer vision methods, and justify design choices in written and oral communication.
4. Apply computer vision techniques to a new problem by framing a measurable question, choosing a suitable model, data, and baseline, and answering the question with honest evaluation.

Course Components

Lectures meet in person, Mon/Wed 3:30–4:50pm in SH 174. Two background topics, Image Formation and Intro to Learning, are not lectured: read their slides before the class listed on the Schedule. Two class meetings are TA-led project labs instead. Slides and readings are posted on Canvas.

Problem sets. Three problem sets, each released at the lecture that unlocks it:

• P1: Images, filtering, and color
• P2: Local features and matching
• P3: CNN recognition

Each has three graded parts, weighted the same way every time:

Reports use patterns you will see in every problem set: results on data you collect yourself, a written prediction before you run an experiment, a parameter sweep instead of a single number, and a hunt for inputs that break the method. Photos you collect are graded on being present, never on quality or on the accuracy you achieve. If a prediction turns out to be wrong, saying so earns full credit.

Midterm. In class, on paper. It covers the material through the lecture before it. The date is on the Schedule.

Final project. A team project that asks and answers an empirical question using open-source vision models. It runs from the first project lab to a video due in finals week, the course’s final-week activity under UCI Senate Regulation A465. See the Project page.

Participation. Short required items, each available from the start of the quarter: the project seed and the team roster. Engagement in class, in the project labs, and on Ed Discussion also counts.

Grading

The final project breakdown is on the Project page. All due dates are on the Schedule.

Grade cutoffs (final percentage, minimum for each grade):

Late Policy

Every deadline has a 2-day grace period. Work submitted during the grace period is accepted without penalty, and you don’t need to ask. Nothing is graded after the grace period ends. The last accepted time for each assignment is shown in Canvas.

The one exception is the final project presentation slides, which have no grace period because you present them in class. Problem Set 1 has a longer grace period for students who add the course late.

The grace period is meant to cover the ordinary surprises of a quarter: illness, travel, interviews, clustered deadlines, computer problems. If a serious illness or emergency lasts beyond it, contact the instructor as soon as you can.

Use of AI Tools

Using AI coding assistants (for example ChatGPT or Claude) is permitted in this course.

Every problem set report ends with a required AI-use disclosure:

• which tools you used,
• roughly what you asked them, and
• at least one specific thing the tool got wrong or misled you about, and how you caught it.

The disclosure is graded on specificity, because noticing where a confident tool is wrong is the actual professional skill.

What you submit must still reflect your own understanding. Your predictions, analyses, and conclusions must be your own. Presenting AI-generated work as your own thinking, or using AI tools without disclosing them, is academic dishonesty.

Academic Integrity

You are expected to pursue your studies with honesty and integrity, and to follow the UCI Academic Integrity Policy. Plagiarism is presenting the ideas, words, or work of another as your own without proper acknowledgment. It includes, but is not limited to:

• submitting work, in part or in whole, that was completed by another person;
• failing to cite sources for ideas, facts, code, or conclusions taken from others;
• closely paraphrasing or substantially reproducing someone else’s work;
• submitting work purchased or obtained from outside sources or services.

Problem set code and reports are individual work. You may discuss ideas with classmates, but do not share code or post solution code on Ed Discussion. The final project builds on open-source code by design. You must separate your team’s contribution from what the original repository already did (see the Project page).

Suspected academic dishonesty is handled under UCI policy. Penalties range from a zero on the assignment to failing the course, and cases may be reported to the Office of Academic Integrity & Student Conduct.

Communication

Ed Discussion is the main channel for course questions: concepts, assignments, and logistics. Posting there lets the whole course staff and your classmates answer faster, and everyone benefits from the answer. The course staff aim to respond within 24 hours on weekdays.

Email is for private matters, such as grading concerns or personal circumstances. Email from your UCI address, and expect a reply within 24–48 hours on business days.

In class, ask questions at any time. Arrive on time, use laptops and phones only for course activities, and keep discussion respectful. Disruptive or dismissive behavior will be addressed under UCI policy.

Materials and Technology

Readings (both free online):

• Computer Vision: Algorithms and Applications, 2nd ed., Richard Szeliski (2022). The primary text.
• Foundations of Computer Vision, Antonio Torralba, Phillip Isola, and William T. Freeman (2024). Supplementary.

Links are on the Resources page.

Technology:

• A laptop that can run Python 3.11. Problem sets 1 and 2 run on the CPU of any laptop.
• Problem set 3 and most final projects need a GPU. The free tier of Google Colab is enough, and every assignment is sized to run there. Final projects should be selected to match your available compute. If unsure, discuss with the instructional staff.
• A reliable internet connection.

Accommodations

Students who need accommodations should register with the UCI Disability Services Center. Share your accommodation letter with the instructor as early in the quarter as possible, and at least a week before the midterm.

Student Resources

Campus resources

• UCI Libraries
• Center for Excellence in Writing and Communication
• Learning & Academic Resource Center (LARC)
• UCI Disability Services Center
• Office of Information Technology (OIT)
• UCI Counseling Center
• Student Affairs: Wellness, Health, and Counseling
• Office of Academic Integrity & Student Conduct

Basic needs

If you have trouble securing enough food or housing, and you believe this may affect your performance in the course, please contact the Dean of Students (949-824-5181) for support. You can also get help from:

• Student Outreach and Retention Center (SOAR)
• UCI Basic Needs Center, at the FRESH Basic Needs Hub, 4079 Mesa Rd

Sexual violence and sexual harassment

Title IX prohibits gender discrimination, including sexual harassment, domestic and dating violence, sexual assault, and stalking. If you have experienced sexual harassment or sexual violence, you can get confidential support and advocacy from the Campus Advocacy Resources & Education (CARE) office, by phone at (949) 824-7273 or by email at care@uci.edu.

You can report incidents through the Office of Equal Opportunity and Diversity. OEOD also lists sexual violence resources that are available even when campus is closed.

Under the UC Policy on Sexual Violence and Sexual Harassment, faculty and TAs must notify the Title IX Office if they learn that a student has experienced sexual violence or sexual harassment.

Acknowledgements

The problem sets descend from Georgia Tech CS 4476/6476 and, before that, Brown CS 143 by James Hays, with contributions from Samarth Brahmbhatt, John Lambert, Judy Hoffman, Viraj Prabhu, Mitch Donley, and Vijay Upadhya. Lecture materials draw on slides by Devi Parikh, Frank Dellaert, Kristen Grauman, David Fouhey, James Hays, Derek Hoiem, and Svetlana Lazebnik. Each slide set and assignment carries its own acknowledgements. You may reuse these materials for academic or research purposes, but please keep all acknowledgements.