Comparing Different Generative Adversarial Networks for Image Generation

Nakov, Dimitar (author), Žabkar, Jure (mentor)

The thesis focuses on exploring the use of various types of Generative Adversarial Networks (GANs) for image generation. We analyze and compare different types of GANs, evaluating their effectiveness in generating high-quality images. Our approach includes a systematic comparison of architectures, implementation strategies, and the quality of generated images. The goal is to provide insight into the advantages and limitations of individual GAN models and to guide future research in this area, contributing to the development of artificial intelligence. Through comprehensive analysis and comparison of different GAN models, we aim to highlight their role in image generation and potential applications in various industries, such as art, entertainment, medicine, and more.

• Language: English
• Keywords: GAN, Artificial Intelligence
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2024
• PID: 20.500.12556/RUL-161308
• COBISS.SI-ID: 211592963
• Publication date in RUL: 09.09.2024

Secondary language

• Language: Slovenian
• Title: Primerjava različnih izvedb generativnih nasprotniških mrež za generiranje slik

Abstract

Diplomsko delo se osredotoča na raziskovanje uporabe različnih vrst Generativnih Nasprotnih Mrež (GANs) za generiranje slik. Analiziramo in primerjamo različne vrste GANs ter ocenjujemo njihovo učinkovitost pri generiranju visoko kakovostnih slik. Naš pristop vključuje sistematično primerjanje arhitektur, implementacijskih strategij in kakovosti ustvarjenih slik. Cilj je ponuditi vpogled v prednosti in omejitve posameznih modelov GANs ter voditi prihodnje raziskave na tem področju, prispevajoč k razvoju umetne inteligence.

• Keywords: GAN, Umetna Intelegenca