The aim of the course is to provide participants with a comprehensive understanding of the principles of machine learning and image processing, while introducing them to the most popular practical applications through concrete, real-world examples. During the training, participants will gain not only theoretical knowledge but also practical skills: they will learn how to use modern tools and methods to create simple image processing models for their own projects. The workshop is designed to enable participants to confidently apply their newly acquired knowledge and to experiment independently in the field of image processing.

Prerequisite knowledge

The course is intermediate level in the field of AI and assumes basic programming knowledge.
(Familiarity with the Python programming language is an advantage.)"

Topics

This course consists of 8 modules, through which participants gradually learn the principles of machine learning and image processing, as well as how to build, evaluate, and interpret models. The program aims to provide both theoretical knowledge and practical skills, enabling participants to independently develop image processing models.

Module 1: Overview of Machine Learning
Participants will learn about the main branches of machine learning: supervised, unsupervised, and reinforcement learning. Key model types and methods will be introduced, along with an overview of underfitting and overfitting, and the role of regularization in improving model reliability.

Module 2: Datasets and Their Challenges
This module addresses common issues with ML datasets, including labeling and data quality. The importance of bias mitigation and fairness will be discussed, as well as cross-validation techniques and model robustness. Data augmentation techniques used in image processing will also be covered.

Module 3: Introduction to Convolutional Neural Networks
Participants will learn how convolutions work, the concepts of stride and padding, and the role of pooling and dropout layers. Classic architectures such as LeNet, AlexNet, VGG, and ResNet will be presented, along with practical interpretation of feature maps.

Module 4: Transfer Learning
We will explore why transfer learning works well, how to perform head replacement, freezing, and partial fine-tuning, and their practical applications.

Module 5: Evaluating Image Processing Models
Participants will learn the characteristics of classification, object detection, and segmentation tasks. Evaluation metrics for classification (ROC, PR curves, confusion matrix) and segmentation metrics (IoU, Dice score) will be introduced.

Module 6: Basics of Object Detection and Localization
This module covers the concept of bounding boxes and the operation of one-stage and two-stage detectors. The popular YOLO family will be presented, along with practical application examples.

Module 7: Segmentation
Participants will learn the differences between semantic and instance segmentation. U-Net and Mask R-CNN models will be introduced, along with typical applications in medical and industrial domains.

Module 8: Explaining Models
The final module focuses on Explainable AI (XAI) techniques, presenting their goals and applications. Integrated gradient-based methods will be discussed in detail, showing how model decisions can be made transparent.

Numbe of hours

8

Group size

10-20

Location

computer lab

The contact person can provide further information about the training and its conditions.