Med prenosom in generacijo obraznega stila

MATEK, SERGEJ (author), Solina, Franc (mentor), Korenič Tratnik, Sebastian (co-mentor)

Dandanes je mnogo stvari v vsakdanjem življenju generiranih z uporabo umetne inteligence, ne da bi se zavedali ali zavedali, kako deluje. Poglobili se bomo v analizo in razvoj tehnologije družine StyleGAN, generacijo fotografij in videov z njo in uporabo rešitev, ki temeljijo na StyleGAN tehnologiji. Uporabljeni bodo odprtokodni modeli GAN, ki jih lahko najdemo v repozitorijih Nvidie z vnaprej naučenimi modeli.

• Language: Slovenian
• Keywords: umetna inteligenca, nevronske mreže, GAN, AI, StyleGAN
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2023
• PID: 20.500.12556/RUL-149716
• COBISS.SI-ID: 165604611
• Publication date in RUL: 08.09.2023

Secondary language

• Language: English
• Title: Between the transfer and generation of facial style

Abstract

Nowadays many things in everyday life are generated using artificial intelligence without most of us knowing or knowing how it works. We will delve into analyzing the development of the StyleGAN family, the generation of images, videos and existing solutions based on StyleGAN technology. We will use the open-source GAN models found in Nvidia's repositories using pre-trained models

• Keywords: artificial intelligence, neural networks, GAN, AI, StyleGAN