Syllabus — Intelligent Vision Systems
7th Semester | BSES, Section A | Fall 2026
Course Overview
| Course Title | Intelligent Vision Systems (Applied Computer Vision) |
| Programme / Semester | Bachelor of Science in Embedded Systems (BSES) — 7th Semester, Section A |
| Semester Duration | 31 August – 14 December 2026 (16 weekly sessions) |
| Class Meeting Times | 11:00 AM–2:00 PM, Monday (3 h = 1.5 h theory + 1.5 h laboratory) |
| Location | E1-404 Lab |
Prerequisites
- Python programming
- Basic linear algebra & probability
Platform: Raspberry Pi 5 (+ AI HAT / Hailo, Pi Camera, ESP32-CAM); Python, OpenCV, PyTorch/TensorFlow, Ultralytics YOLO; Google Colab for training.
Course Website: this site
Instructor
Dr. Muhammad Kazim — Assistant Professor
Department of Intelligent Systems, University of Lahore
- Preferred form of address: Please call me Kazim.
- Office: TBA
- Office hours: By appointment
- Email: muhammad.kazim@is.uol.edu.pk
Course Description
Intelligent Vision Systems (Applied Computer Vision) is the capstone vision course for Embedded Systems students. It develops the complete machine-vision pipeline — from photons striking a sensor to a decision produced within the memory, power, and latency budget of a constrained device.
The course begins with image formation and classical image processing, then moves through feature detection, multi-view geometry, and motion analysis, and finishes with modern deep-learning models for classification, detection, and segmentation. A recurring theme is deployment: taking an algorithm or a trained model and making it run in real time on the Raspberry Pi 5, optionally accelerated with the AI HAT/Hailo NPU or fed by a networked ESP32-CAM. The course is project-driven: each 3-hour Monday session pairs 1.5 hours of theory with 1.5 hours of supervised laboratory work, and a capstone project asks every student to build and demonstrate a real-time vision application end-to-end on the Raspberry Pi 5.
Learning Outcomes
On successful completion, a student will be able to:
- CLO1. Explain the fundamentals of image formation, sensing, representation, colour, and processing.
- CLO2. Apply filtering, convolution, frequency-domain analysis, edge detection, and Hough methods.
- CLO3. Detect, describe, and match visual features and estimate geometric transforms robustly.
- CLO4. Apply camera geometry, calibration, stereo vision, and motion estimation for 3-D and dynamic scenes.
- CLO5. Implement classical recognition/detection (e.g. HOG, Haar, SIFT) and evaluate them.
- CLO6. Apply machine and deep learning to image classification, object detection, tracking, and segmentation.
- CLO7. Develop, deploy, and evaluate real-time computer-vision pipelines on embedded hardware (Raspberry Pi 5).
Textbooks & References
| Use | Title | Author(s) · Edition | Notes |
|---|---|---|---|
| Primary | Computer Vision: Algorithms and Applications | R. Szeliski · 2nd ed. | Free at szeliski.org/Book; core reading for weeks 1–11 (image formation through stereo). |
| Primary | Digital Image Processing | Gonzalez & Woods · 4th ed. | Best treatment of weeks 2–9 (filtering, frequency domain, morphology, segmentation). |
| Primary | Deep Learning | Goodfellow, Bengio & Courville | Free at deeplearningbook.org; reference for weeks 12–13. |
| Practical | OpenCV 5.0 documentation & 4→5 migration guide | OpenCV team | API reference for every lab; deep dive in week 14. |
Open university courses: Cornell CS5670 (cs.cornell.edu/courses/cs5670) · UW CSE455 (courses.cs.washington.edu/courses/cse455) · UMich EECS 498, Deep Learning for CV (web.eecs.umich.edu/~justincj/teaching/eecs498) · Stanford CS231n (cs231n.github.io).
Key research papers (implementation labs): Vaswani et al. (2017), “Attention Is All You Need” — arxiv.org/abs/1706.03762 (Lab 15: Transformer from scratch) · Ronneberger et al. (2015), “U-Net” — arxiv.org/abs/1505.04597 (Lab 13: U-Net from scratch).
Assessment & Grading
Each 3-hour Monday session pairs a 1.5-hour theory lecture with a 1.5-hour supervised laboratory on the Raspberry Pi 5.
| Assessment | Marks | Coverage |
|---|---|---|
| Final Exam | 40 | Cumulative; weighted toward recognition, deep learning & embedded deployment (weeks 10–16). |
| Midterm Exam | 20 | Week 8 — image formation, processing, features, matching & RANSAC (weeks 1–8). |
| Lab Project (capstone) | 30 | End-to-end vision system deployed on the Raspberry Pi 5 + demo & report. |
| Quizzes / Assignments | 10 | Four assignments + periodic quizzes across the semester. |
| Total | 100 |
The Lab Project (30) is the capstone system demonstrated live on target hardware in Week 16; the four assignments and quizzes together carry 10 marks. Weekly lab deliverables are checkpoints toward the capstone and count within lab participation.
Grade Scale
| A | A− | B+ | B | B− | C+ | C | C− | D | F |
|---|---|---|---|---|---|---|---|---|---|
| ≥90 | 86–89 | 82–85 | 78–81 | 74–77 | 70–73 | 66–69 | 62–65 | 50–61 | <50 |
Adjust cut-offs to your institution’s regulations.
Course Policies
Late Work
Each student has a budget of 4 late days for the term, used in whole-day units, at most 2 per artefact. After the budget is spent, the maximum score drops 15% per day for up to 3 days, then reaches zero. Exams and the final presentation cannot use late days.
Collaboration & Academic Integrity
Discussing concepts is encouraged. The code and written answers you submit must be your own (or your team’s for the project). Cite any external code, dataset, or AI assistance you use. Copying solutions, sharing code between teams, or submitting generated text as your own analysis are integrity violations and are reported per university policy.
AI Tools
You may use AI assistants for debugging, boilerplate, and explanation. You may not use them to generate whole lab reports or exam answers. Add a one-line “AI use” note to each submission describing what you used them for.
Attendance
Lecture attendance is expected; labs are mandatory because equipment and check-offs happen in the room. Notify the instructor before a missed lab to arrange a make-up slot.
Accommodations & Well-being
Students with approved accommodations should share their letter in the first two weeks so arrangements can be made discreetly. If you are struggling, contact the instructor early — support is easier to arrange before a deadline than after.
Weekly Topic Map
The full lecture-by-lecture breakdown with readings and deadlines is on the Schedule page; lab details are on the Labs page.
| Weeks | Block | Milestones |
|---|---|---|
| 1–2 | Introduction & the edge pipeline; image formation I & II; digital fundamentals | Lab 1–2 (RPi 5 bring-up, OpenCV 5.0, camera stack) |
| 3–4 | Point operations, thresholding, spatial filtering I & II | Lab 3–4 · Assignment 1 out/due |
| 5 | Frequency-domain processing & pyramids | Lab 5 |
| 6–7 | Restoration & morphology; edges/contours & local features | Lab 6–7 · Assignment 2 out/due |
| 8 | Descriptors, matching & RANSAC | Lab 8 · Midterm Exam (20) |
| 9 | Segmentation I & II | Lab 9 · Project proposal due |
| 10 | Calibration & stereo | Lab 10 |
| 11 | Optical flow & tracking | Lab 11 · Assignment 3 out |
| 12 | Classical recognition & neural networks | Lab 12 · Assignment 3 due |
| 13 | CNNs, segmentation networks & reading research papers | Lab 13 — U-Net from scratch · Assignment 4 out |
| 14 | Edge vision & the OpenCV 5.0 deep dive | Lab 14 · Assignment 4 due |
| 15 | Vision transformers, advanced topics & review | Lab 15 — Transformer from scratch · Project report due |
| 16 | Capstone demos & review | Capstone demos · Final Exam (40, finals week) |
Software & Hardware
Software: Python, OpenCV 5.0, PyTorch/TensorFlow, Ultralytics YOLO (incl. YOLO26); Google Colab for training.
Hardware: Raspberry Pi 5 + AI HAT / Hailo NPU, Pi Camera, and an ESP32-CAM as a networked sensor — all labs and the capstone target this platform.
Tools & platforms: Ultralytics YOLO (docs.ultralytics.com) · Raspberry Pi AI / Hailo (raspberrypi.com/documentation/computers/ai.html) · OpenCV 5.0 (docs.opencv.org) · Roboflow (roboflow.com) · Google Colab (colab.research.google.com).
Model and tool versions move quickly — reconfirm YOLO26, OpenCV 5.0, Hailo and library versions before each offering.
Setup guidance is on the Tutorials page.