Unit 753: Computer Vision Fundamentals and Applications

1. Introduction to Computer Vision

• Definition and scope of computer vision
• Key applications: surveillance, autonomous vehicles, medical imaging, robotics
• Importance of computational understanding of images and videos
• Challenges in computer vision

2. Image Processing Fundamentals

2.1 Image Representation

• Pixels and image formats
• Grayscale vs. color images

2.2 Color Models

• RGB, HSV, CMYK, and other color spaces

2.3 Image Enhancement

• Contrast adjustment, histogram equalization
• Noise reduction techniques

2.4 Filtering

• Spatial filters: smoothing, sharpening
• Frequency domain filtering basics

2.5 Image Transformations

• Geometric transformations: translation, rotation, scaling
• Fourier transform essentials

3. Image Segmentation

• Purpose and importance of segmentation
• Thresholding methods
• Edge-based segmentation
• Region-based segmentation
• Clustering approaches (e.g., K-means)
• Advanced segmentation: Watershed, Graph cuts

4. Object Detection and Recognition

4.1 Feature Extraction

• Keypoint detectors (SIFT, SURF, ORB)
• Descriptors and feature matching

4.2 Object Localization

• Sliding window approach
• Region proposal methods

4.3 Classification Algorithms

• Traditional machine learning: SVM, Random Forest
• Introduction to deep learning-based detection

5. Convolutional Neural Networks (CNNs)

• CNN architecture: convolutional layers, pooling, fully connected layers
• Training CNNs for image tasks
• Popular CNN models (LeNet, AlexNet, VGG, ResNet)
• Applications in feature extraction and hierarchical learning

6. Transfer Learning in Computer Vision

• Concept and benefits of transfer learning
• Using pre-trained models (e.g., ImageNet models)
• Fine-tuning techniques
• Case studies and practical examples

7. Face Recognition and Biometrics

• Face detection techniques (Haar cascades, MTCNN)
• Face alignment and normalization
• Feature extraction for faces
• Recognition algorithms: Eigenfaces, Fisherfaces, deep learning approaches
• Introduction to other biometric modalities

8. Object Tracking

• Problem definition and challenges
• Tracking algorithms overview
• Optical flow methods
• Kalman and particle filters
• Deep learning approaches in tracking

9. Image Captioning and Visual Question Answering

• Overview of multimodal AI tasks
• Image captioning architectures (CNN + RNN)
• Visual question answering models
• Datasets and evaluation metrics

10. Ethical Considerations in Computer Vision

• Privacy concerns and surveillance
• Algorithmic bias and fairness
• Societal impact and responsible AI use
• Legal and regulatory perspectives