Objectives and competences
The goal of the course is for the students to understand the basics of deep networks in the field of computer vision and be able to apply them in practice.
Content (Syllabus outline)
• Uvod: opredelitev osnovnih pojmov
• Barvne in geometrijske transformacije: barvni prostori in osnovne geometrijske transformacije
• Konvolucija in linearni operatorji: detektorji robov, primeri detektorjev robov
• Segmentacija slik
• Globoke nevronske mreže in računalniški vid: sloji, učni algoritmi, prenosno učenje
• Priprava učnih podatkov: predobdelava in izboljšava podatkov (augmentacija), odstranjevanje šuma
• Interpretacija globokih modelov: podprileganje in nadprileganje, vrednotenje modelov globokih mrež.
• Primeri aplikacij globokih mrež: klasifikacija, detekcija, segmentacija.
Learning and teaching methods
Lectures,
Lab work.
Intended learning outcomes - knowledge and understanding
Learning outcome: identify computer vision problems, and select and use appropriate deep network models
Learning outcome: Understand the fundamental building blocks and the architecture of deep neural network models
Learning outcome: Apply deep neural network models to solve computer vision problems
Learning outcome: Reuse trained deep neural network models to solve new computer vision problems
Learning outcome: Analyze and compare solutions for computer vision problems.
Readings
F. Chollet, Deep Learning with Python, druga izdaja, Manning, 2021.
E. R. Davies, Computer Vision: Principles, Algorithms, Applications, Learning, peta izdaja, Academic Press, 2017.
R. Shanmugamani, Deep Learning for Computer Vision, Packt Publishing, 2018.
Literatura v slovenskem jeziku za tematiko, ki pokriva celoten predmet ne obstaja.
