Computer Vision

Course objectives

GENERAL OBJECTIVES The course aims to introduce the student to the fundamental concepts of artificial vision and to the construction of autonomous systems of interpretation and reconstruction of a scene through images and video. The course deals with basic elements of projective and epipolar geometry, methods for 3D vision and vision based on multiple views, and methods for metric reconstruction and image and video interpretation methods. Furthermore the course illustrates the main techniques for the recognition and segmentation of images and videos based on machine learning. SPECIFIC OBJECTIVES Knowledge and Understanding The course stimulates students' curiosity towards new methodologies for the analysis and generation of images and video. The student learns new concepts that allow him to acquire a basic knowledge of computational vision. Apply Knowledge and Understanding Students deepen and learn programming languages ??to apply the acquired knowledge. In particular they deepen the Python language and learn Tensorflow. The latter offers students the possibility of programming deep learning applications. They use this brand new technology to make a project to recognize specific elements in images and videos. Critical and Judgment skills The student acquires the ability to distinguish between what he can achieve with the tools he/she has learned, such as generating images or recognizing objects using deep learning techniques, and what is actually required for the realization of an automatic vision system. In this way she/he is able to elaborate a critical judgment on the vision systems available to the state of art and to assess what can actually be achieved and what requires further progress in research. Communication skills The realization of the project, as part of the exam program, requires the student to work and give a contribution within a small work group. This together with the solution of exercises in the classroom, and to the discussions on the most interesting topics it stimulates the student's communication skills. Learning ability In addition to the classic learning skills provided by the theoretical study of the teaching material, the course development methods, in particular the project activities, stimulate the student to the self-study of some topics presented in the course, to group work, and to the application concrete knowledge and techniques learned during the course.

Channel 1
IRENE AMERINI Lecturers' profile

Program - Frequency - Exams

Course program
1. Image Processing Image Formation and Filtering Frequency analysis Feature Detection and Matching Motion and Optical flow 2. Multiview Geometry Cameras, Multiple Views Projective geometry and camera geometry 3. Deep Learning for Computer Vision Deep Learning for computer vision and basic architectures Object classification and detection, semantic segmentation and instance segmentation Sequence models: RNN, transformers in vision and attention model, action recognition Generative models Monocular depth estimation 4. Computer Vision Security and Forensics Dataset Bias and Adversarial examples Multimedia forensics and deepfake detection
Prerequisites
Elements of programming (Python), Machine Learning
Books
Suggested (not mandatory) - Computer Vision and Image Processing: «Computer Vision: Algorithms and Applications», Richard Szeliski , 2022 (available for free at: https://szeliski.org/Book/) - MultiView Geometry: “MultiView Geometry in Computer Vision”, Richard Hartley & Andrew Zisserman, Cambridge ed. - Understanding Deep Learning. J.D. Prince, https://udlbook.github.io/udlbook/ - Deep Learning with PyTorch, Eli Stevens, Luca Antiga, Thomas Viehmann - Deep learning: «Deep Learning», Ian Goodfellow , Yoshua Bengio , Aaron Courville (available at: https://www.deeplearningbook.org/) - Scientific papers and articles discussed during the course
Teaching mode
Lectures and tutoring sessions at the computer
Frequency
non-mandatory attendance
Exam mode
● The exam covers the different sections of the course (1-4) ● The assessment of the exam consist of a project + project presentation (worth 1/2) and a final written exam (worth 1/2). ● Final project: Algorithms, objectives and topics for the final project may be freely chosen (a list of topics will be given at the half of the course). ● It requires a project abstract to be presented at the end of the course and approved by the instructors ● Groups from 1 to 3 people
Lesson mode
Lectures and tutoring sessions at the computer
IRENE AMERINI Lecturers' profile

Program - Frequency - Exams

Course program
1. Image Processing Image Formation and Filtering Frequency analysis Feature Detection and Matching Motion and Optical flow 2. Multiview Geometry Cameras, Multiple Views Projective geometry and camera geometry 3. Deep Learning for Computer Vision Deep Learning for computer vision and basic architectures Object classification and detection, semantic segmentation and instance segmentation Sequence models: RNN, transformers in vision and attention model, action recognition Generative models Monocular depth estimation 4. Computer Vision Security and Forensics Dataset Bias and Adversarial examples Multimedia forensics and deepfake detection
Prerequisites
Elements of programming (Python), Machine Learning
Books
Suggested (not mandatory) - Computer Vision and Image Processing: «Computer Vision: Algorithms and Applications», Richard Szeliski , 2022 (available for free at: https://szeliski.org/Book/) - MultiView Geometry: “MultiView Geometry in Computer Vision”, Richard Hartley & Andrew Zisserman, Cambridge ed. - Understanding Deep Learning. J.D. Prince, https://udlbook.github.io/udlbook/ - Deep Learning with PyTorch, Eli Stevens, Luca Antiga, Thomas Viehmann - Deep learning: «Deep Learning», Ian Goodfellow , Yoshua Bengio , Aaron Courville (available at: https://www.deeplearningbook.org/) - Scientific papers and articles discussed during the course
Teaching mode
Lectures and tutoring sessions at the computer
Frequency
non-mandatory attendance
Exam mode
● The exam covers the different sections of the course (1-4) ● The assessment of the exam consist of a project + project presentation (worth 1/2) and a final written exam (worth 1/2). ● Final project: Algorithms, objectives and topics for the final project may be freely chosen (a list of topics will be given at the half of the course). ● It requires a project abstract to be presented at the end of the course and approved by the instructors ● Groups from 1 to 3 people
Lesson mode
Lectures and tutoring sessions at the computer
  • Lesson code1052229
  • Academic year2024/2025
  • CourseArtificial Intelligence and Robotics
  • CurriculumSingle curriculum
  • Year1st year
  • Semester2nd semester
  • SSDING-INF/05
  • CFU6
  • Subject areaIngegneria informatica