Computer Vision

General Information

• University: University of Kurdistan Hewlêr
• Department: Computer Engineering Dept.
• My Status: VIsiting Lecturer
• Level: MSc
• Year: 2018-2020

Course Description

This course provides a comprehensive introduction to computer vision, covering both theoretical foundations and practical applications. Students will learn the fundamental principles of visual information processing, image analysis, and pattern recognition techniques used in modern computer vision systems.

The curriculum emphasizes problem-based learning, where students work on real-world computer vision challenges in ubiquitous computing and entertainment applications. Through hands-on projects and practical exercises, students will develop the skills necessary to design and implement computer vision solutions for various domains including robotics, autonomous vehicles, medical imaging, and multimedia applications.

The course combines mathematical theory with practical implementation, enabling students to understand both the underlying principles and the practical challenges of applying computer vision technology to real-world problems.

Prerequisites

• Linear Algebra and Calculus
• Probability and Statistics
• Programming in Python (or equivalent)
• Machine Learning fundamentals
• Digital Image Processing basics (recommended)

Course Objectives

Upon completion of this course, students will be able to:

• Understand the fundamental principles and mathematical foundations of computer vision.
• Implement and evaluate computer vision algorithms for image processing and analysis.
• Design and train deep learning models for visual recognition tasks.
• Apply computer vision techniques to solve real-world problems in various domains.
• Analyze and optimize the performance of computer vision systems.
• Understand the limitations and challenges of current computer vision technology.

Course Outline

Module 1: Introduction to Computer Vision

• Computer vision fundamentals and applications
• Image formation and camera models
• Image representation and digital image processing basics
• Computer vision pipeline and workflow
• Current trends and challenges in computer vision

Module 2: Image Classification Pipeline

• Traditional image classification methods
• Feature extraction and representation
• Machine learning approaches for classification
• Evaluation metrics and performance assessment
• Image preprocessing and augmentation techniques

Module 3: Neural Networks and Deep Learning

• Artificial neural network fundamentals
• Backpropagation algorithm and training
• Loss functions and optimization techniques
• Neural network architectures for vision tasks
• Training strategies and regularization methods

Module 4: Convolutional Neural Networks (CNNs)

• CNN architecture and components
• Convolutional layers and feature maps
• Pooling and normalization techniques
• CNN training and optimization
• Transfer learning and fine-tuning strategies

Module 5: Advanced CNN Architectures

• AlexNet, VGG, and ResNet architectures
• Inception networks and GoogLeNet
• Modern CNN architectures (EfficientNet, etc.)
• Architecture design principles and best practices
• Performance comparison and selection criteria

Module 6: Object Detection and Localization

• Sliding window and region proposal methods
• R-CNN family of algorithms
• YOLO and single-stage detectors
• Object detection evaluation metrics
• Real-time object detection systems

Module 7: Image Segmentation

• Semantic segmentation techniques
• Instance segmentation methods
• U-Net and encoder-decoder architectures
• Mask R-CNN and advanced segmentation models
• Segmentation applications and challenges

Module 8: Recurrent Neural Networks for Vision

• RNN fundamentals and LSTM networks
• Video analysis and temporal modeling
• Action recognition and video classification
• Sequence-to-sequence models for vision
• Attention mechanisms in vision tasks

Module 9: Feature Detection and Matching

• Corner detection and feature points
• SIFT, SURF, and ORB algorithms
• Feature matching and homography estimation
• Image stitching and panorama creation
• Structure from motion techniques

Module 10: Visual Understanding and Interpretation

• Visual attention and saliency detection
• Scene understanding and context analysis
• Visual question answering systems
• Image captioning and visual storytelling
• Interpretability and explainable AI in vision

Module 11: Generative Models in Computer Vision

• Generative Adversarial Networks (GANs)
• Variational Autoencoders (VAEs)
• Image generation and synthesis
• Style transfer and image manipulation
• Deepfake detection and prevention

Module 12: Computer Vision Applications

• Medical imaging and diagnosis
• Autonomous vehicles and robotics
• Augmented and virtual reality
• Surveillance and security systems
• Industrial inspection and quality control

Textbooks

• [Recommended] “Computer Vision: Algorithms and Applications” by Richard Szeliski
• [Optional] “Deep Learning for Computer Vision” by Adrian Rosebrock

Assessment

• Midterm Exam (20%)
• Activities and Projects (30%)
  • Programming Assignments (15%)
  • Computer Vision Projects (15%)
• Final Exam (50%)