Insegnamento a.a. 2026-2027

Convolutional neural networks are mature, flexible, and powerful non-linear data-driven models that have successfully been applied to solve complex tasks in science and engineering. The advent of the deep learning paradigm, i.e., the use of neural networks to simultaneously learn an optimal data representation and the classification model, has further the data-driven paradigm. These topics will be described in the course according to the following detailed program: 

• Introduction to Computer Vision and basics of digital images

• Basics of image transformations and image filtering

• Image Classification with Linear Classifiers

• Convolutional Neural Networks for Image Classification

• CNNs Architectures

• Advanced Deep Learning architectures

• Object Detection, Image Segmentation

• Techniques for Visual Data Visualization and Interpretation

• Unsupervised and Self-supervised Learning

• Generative Models

• Emerging Topics in Vision: Video Understanding, 3D Perception, Multimodal Models

• Identify the right CNN architecture to solve different visual recognition problems 

• Recognize the best practices, leveraging the most popular dropout, data augmentation 

• Describe and get inspiration from the most successful Deep Learning architectures 

• Explain the most successful Computer Vision applications to be solved by Deep Learning models 

• Illustrate complex techniques beyond the fundamental ones presented during lectures 

• Analyze a specific Computer Vision problem and find which model best solves the task at hand 
• Use fundamental deep learning algorithms for Computer Vision autonomously 
• Compare the various models and find the most relevant to be applied in the specific problem 
• Examine the selected model in order to balance performance, computational complexity and overfitting 
• Discuss the pros and cons of different Computer Vision techniques for a specific problem 
• Develop new pipelines adapting to the specific problem at hand 

• Lectures
• Guest speaker's talks (in class or in distance)
• Practical Exercises
• Collaborative Works / Assignments

The course follows an interactive and hands-on teaching modality with a strong emphasis on practical aspects. On top of the laboratory sessions, customarily held after most lectures, the course leverages project-based learning to enable students to apply the principles covered during lectures to real-world computer vision tasks. 
During Practical Session carefully selected sample codes cover the key components of image analysis, and convolutional neural networks for image classification, segmentation, object recognition, and image generation. Students are encouraged to follow along and experiment with the code to gain a solid grasp of the underpinning concepts. 
Projects are assigned to groups to foster a deeper understanding of the subject. The students are divided into teams, and will phase two step-projects. The first phase, which will take place during the first half of the course, is meant to teach the students how to use CNN models for solving a basic visual recognition task. In the second phase, students are invited to choose a specific computer vision problem to be solved by advanced deep learning models. The projects need to be diverse among the teams, challenging, and relevant to current real-world applications. 
During the project development, students are expected to take advantage of the methods and skills presented during lectures for solving their specific task. At the end of the course, each team presents their projects to the entire class. A poster session will also take place where students have the change to discuss their project with their peers in front of a nice poster summarizing their work, in a "conference style". The latter fosters a collaborative learning environment where teams can learn from each other's successes and challenges. 

• Assignment / Project work (60%): Students work in teams of 3 on a deep learning project related to computer vision. The assignment consists of the design, implementation, training, validation, and critical evaluation of a deep learning model for a visual recognition task. The project is assessed through a written report, a final oral presentation (in class), and a poster session (in class) (simulates a real scientific conference setting, where each group presents and discusses the project results through a poster). 
  This assessment verifies that students are able to apply the theoretical concepts learned in the course, select appropriate neural network architectures, manage datasets, train and evaluate models, interpret experimental results, and communicate technical choices and outcomes clearly.

  The project grade is expressed on a scale of 0–21 points and is composed as follows:

  • Presentation: 20% (4 points)
  • Report: 60% (12 points)
  • Poster session: 20% (4 points)
  • 1 additional point for exceptional projects
• Written Exam (40%): The written exam consists of 6–10 open and/or closed questions, lasting 1 hour. Students are required to use pen, paper, and calculator. The exam is graded on a scale of 0–10 points.
  The written exam verifies students’ knowledge and understanding of the main concepts of deep learning for computer vision, including neural network architectures, training procedures, optimization, regularization, evaluation metrics, and the ability to reason about methodological choices and expected model behavior.
• Collaborative work / in-class exercises, bonus up to 1 point: During the course, students may take part in collaborative exercises carried out in class, individually or in small groups. These activities consist of practical exercises on deep learning models, computer vision tasks, training procedures, and interpretation of results.
  This activity contributes to verifying students’ ability to apply course concepts to concrete problems, discuss methodological choices, and develop practical problem-solving skills. A bonus of up to 1 additional point may be added to the final grade.

• Written Exam (100%): The written exam consists of 12–20 open and/or closed questions, lasting 1 hour and 40 minutes. Students are required to use pen, paper, and calculator. The exam is graded on a scale of 0–31 points.
  The exam assesses the full set of expected learning outcomes for the course. It verifies students’ theoretical knowledge of deep learning methods for computer vision, their understanding of neural network architectures and training strategies, and their ability to analyze, compare, and select appropriate models and evaluation procedures for computer vision tasks. The open questions also assess students’ capacity to reason critically about experimental results, limitations, and practical design choices.

• Collaborative work / exercises, bonus up to 1 point: Non-attending students may obtain a bonus by completing the same type of exercises proposed during the course, carrying them out independently from home. These exercises focus on the application of deep learning methods to computer vision problems, including model design, training choices, evaluation, and discussion of results.
  This activity contributes to verifying students’ ability to apply theoretical knowledge to practical problems and to reason about methodological and experimental choices. A bonus of up to 1 additional point may be added to the final grade.

Course Material

The teaching materials made available to students include:

• Slides used during lectures, providing an overview of key concepts and methods.

• Lecture notes that expand on the topics presented in class and serve as a study reference.

• Practical session notebooks, accompanied by solutions to the exercises, designed to support hands-on learning and independent practice.