Computer Vision & Image Processing Projects

This repository contains the lab assignments for Computer Vision & Image Processing at PolyU.

The projects cover a wide range of topics including fundamental image processing algorithms, geometric computer vision, and deep learning using PyTorch.

📂 Project Structure

Lab 1: Filter Basics & Edge Detection

Implementation of fundamental image processing algorithms from scratch.

• Key Tasks: Manual implementation of Gaussian/Median filters.
• Edge Detection: Canny Edge Detector (Sobel operator, NMS, Hysteresis thresholding).
• Line Detection: Hough Transform implementation.

Lab 2: Keypoints Detection & Matching

Feature extraction and geometric applications using SIFT and ORB.

• Feature Matching: SIFT algorithm with Lowe's ratio test.
• Panorama: Image stitching using Homography matrix estimation.
• Video Stabilization: Removing camera shake from video footage.

Lab 3: Image Classification

Deep learning models for image recognition implemented with PyTorch.

• CNN Implementation: Building a custom CNN for MNIST classification.
• Transfer Learning: Fine-tuning ResNet18 for the Fashion-MNIST dataset.
• Optimization: Learning rate scheduling (MultiStepLR) and data augmentation.

🛠️ Tech Stack

• Language: Python
• Libraries: OpenCV, PyTorch, NumPy, Matplotlib

---

Created for coursework.