ECE4076/5176 Computer Vision

Interactive class notebooks for ECE4076/5176 Computer Vision at Monash University.

This repository contains in-class activities using Jupyter notebooks alongside pre-recorded lectures, paired with laboratory sessions where students implement related concepts. The course covers both classical computer vision and modern deep learning methods.

---

Course Content

Main Weeks

• Week 1: Image handling and basic manipulation, high dimensional signals
• Week 2: Image filtering, Difference of Gaussians, Keypoint Detection, Patch Matching
• Week 3: Invariances and image transformations, asymmetric feature matching
• Week 4: Camera models and homography estimation using RANSAC
• Week 5: Camera projection, vanishing points and lines
• Week 6: Multiple view geometry, space carving
• Week 7: Clustering and Gaussian mixture models
• Week 8: Logistic regression and face detection
• Week 9: Understanding gradient descent
• Week 10: Object detection with AlexNet and ResNets, Attention
• Week 11: Object recognition with RCNNs and YOLO
• Week 12: Image segmentation with U-Nets

Bonus Content

Additional material available in the bonus_content/ folder:

• Vision Transformers
• Variational Autoencoders
• Vector Quantised Variational Autoencoders
• Generative Adversarial Networks
• CLIP (Contrastive Language-Image Pre-training)
• Diffusion Models

---

Installation (5 minutes)

Warning: This installation will take a significant amount of disk space. If your hard drive is full, you may want to consider opening the notebooks in Google Colab instead and installing required packages as needed.

Prerequisites

• Python 3.8+ - Download from python.org
  • Windows users: Check "Add Python to PATH" during installation
• VS Code (recommended) - Download from code.visualstudio.com

Setup Steps

Step 1: Run the installer

Open Terminal/Command Prompt in this folder and run:

# Linux/macOS
chmod +x install.sh
./install.sh

# Windows
install.bat

Step 2: Open in VS Code

• File → Open Folder → Select this ECE4076 folder
• Wait for VS Code to load (~10 seconds)
• Click "Install" when prompted to install recommended extensions

Step 3: Start a notebook

• Click any .ipynb file in the Explorer panel
• Select the ECE4076 kernel when prompted
• You're ready to go!

Alternative: Command Line Only

If you prefer Jupyter in your browser:

# Activate environment
source ece4076_env/bin/activate      # Linux
. ece4076_env/bin/activate           # macOS
ece4076_env\Scripts\activate.bat      # Windows

# Start Jupyter
jupyter notebook

---

VS Code Features

Running Cells

Action	Shortcut
Run cell	Click ▶ or Ctrl+Alt+Enter
Run all cells	View → Run All Cells
Run cells above	Cell menu → Run Above

---

Troubleshooting

"Python is not installed"

Solution: Download and install from python.org

• Windows: Check "Add Python to PATH" during installation
• macOS: Use Homebrew: brew install python3
• Linux: Use package manager: apt install python3 or yum install python3

Verify: python3 --version

"Permission denied" (Linux/macOS)

Error: bash: ./install.sh: Permission denied

Solution:

chmod +x install.sh
./install.sh

Installation hangs or takes too long

Solution: Be patient! PyTorch downloads can take time.

• Check internet connection
• If interrupted, run the script again—it resumes from where it left off
• Ensure you have ~2GB free disk space

Python package issues (especially on Windows)

Solution: Use the conda-based installer instead:

# Linux/macOS
chmod +x install_conda.sh
./install_conda.sh

This creates a self-contained conda environment and avoids common dependency conflicts. Windows users: ensure conda is installed, then run the bash command above in a conda prompt or PowerShell.

"ModuleNotFoundError" when running notebooks

Solution: Make sure you're using the ECE4076 kernel

1. In VS Code: Look for "Select Kernel" button (top right of notebook)
2. Choose "ECE4076 (./ece4076_env/bin/python)"

If not available:

source ece4076_env/bin/activate  # Linux/macOS
python -m ipykernel install --user --name ece4076 --display-name "ECE4076"

"Kernel not found" in VS Code

Solution:

1. Terminal → New Terminal (Ctrl+`)
2. Run:

   source ece4076_env/bin/activate  # Linux/macOS
   python -m ipykernel install --user --name ece4076 --display-name "ECE4076"

3. Reload VS Code (Ctrl+Shift+P → Reload Window)

"ImportError: No module named 'cv2'" or similar

Solution: Activate the environment before running notebooks

source ece4076_env/bin/activate      # Linux
. ece4076_env/bin/activate           # macOS
ece4076_env\Scripts\activate.bat      # Windows
jupyter notebook

Need to reinstall everything

# Linux/macOS
rm -rf ece4076_env
./install.sh

# Windows
Delete the ece4076_env folder manually
Then double-click install.bat

Still stuck?

1. Check the error message—it often contains the solution
2. Search the error online (Gemini, DeepSeek, GPT, etc.)
3. Ask on our forums

---

Preparation

You may want to brush up on Python skills before starting. Check out this Python intro notebook.

---

Contributors & Acknowledgements

• Michael Burke
• Mehrtash Harandi
• Aaron Choong
• Haoyang Jiang
• Week 5 based on an example from Tom Drummond