Παρουσίαση/Προβολή

Διδάσκοντες

Επίκουρος Καθηγητής Ευτύχιος Πρωτοπαπαδάκης

Περιεχόμενο μαθήματος

Lesson 1: Introduction to Machine Learning and Computer Vision

• Overview of machine learning and computer vision
• Image representation and processing
• Key challenges in computer vision

Lesson 2: Probability, Statistics, and Linear Algebra Review

• Probability theory essentials for machine learning
• Statistical concepts and distributions
• Linear algebra concepts for computer vision

Lesson 3: Feature Extraction Techniques

• What is an image, basic properties
• Image feature representation (e.g., color, texture, edges)
• Local feature descriptors (e.g., SIFT, SURF, ORB)
• Bag of Visual Words (BoW) model

Lesson 4: Supervised Learning Algorithms

• Linear regression and regularization techniques
• Support Vector Machines (SVM)
• Decision Trees and Random Forests
• Naïve Bayes
• Performance evaluation metrics

Lesson 5: Neural Networks and Deep Learning

• Feedforward neural networks
• Activation functions and backpropagation
• Stacked autoencoders for classification
• Training deep neural networks

Lesson 6: Unsupervised Learning and Dimensionality Reduction

• K-means clustering
• Principal Component Analysis (PCA)
• t-distributed Stochastic Neighbor Embedding (t-SNE)
• Encoding using SAEs

Lesson 7: Convolutional Neural Networks (CNNs)

• CNN architecture and components
• Transfer learning and fine-tuning in computer vision
• Object detection with CNNs
• Additional performance scores IoU

Lesson 8: Object Detection Techniques

• Introduction to object detection
• One-shot detectors (YOLO, SSD)
• Two-step detectors (R-CNN, Faster R-CNN)

Lesson 9: Semantic Segmentation

• Introduction to semantic segmentation
• Fully Convolutional Networks (FCN)
• U-Net and its variations

Lesson 10: Generative adversarial networks

• Introduction to GANs and their architecture
• Training process and loss functions
• Applications of GANs in image synthesis
• StyleGAN2, BigGAN, and other variations, as well as applications beyond image synthesis, such as data augmentation and domain adaptation.

Lesson 11: Other Computer Vision Techniques

• Object tracking in videos
• Face recognition and emotion detection
• Optical character recognition (OCR)

Lesson 12: Interpretability and Adversarial Robustness in Computer Vision

• Interpreting deep learning models
• Adversarial attacks and defenses
• Model robustness and uncertainty estimation

Lesson 13: Final Project

• Students work on a hands-on computer vision project
• Project presentation and review

Course grade

1. 3 x exercises [up to four person teams] - 25%
   1. Comparing feature extraction techniques and classification approaches
   2. Clustering over deep learning embeddings vs classical feature extraction techniques
   3. Object detection and/or semantic segmentation application
2. Presentation of a case study [up to two person teams] - 25%
   1. Motivation & related work
   2. Dataset description
   3. Proposed methodology
   4. Experimental results
3. Final exam - 50%
   1. Multiple choice questions
   2. No notes, no calculators
   3. 1 hour duration