Primjena generativnih suparničkih mreža na generiranje slika

master's thesis

• Title (english): Application of Generative Adversarial Networks on Image Generation
• Author: Luka Banović
• Mentor: Marko Čupić (mentor)
• Committee member: Marko Čupić (predsjednik povjerenstva)
• Committee member: Jan Šnajder (član povjerenstva)
• Committee member: Domagoj Jakobović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2020-07-08, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Generativne suparničke mreže moćna su klasa dubokih modela i vrlo aktivno područje istraživanja. U ovom radu predstavljamo njihov osnovni koncept te iznosimo ideju iza generativnih modela općenito. Budući da se model u najširem smislu temelji na suradnji između dvije duboke neuronske mreže, na visokoj razini pojašnjavamo kako i one funkcioniraju. Iznosimo i uobičajeno korištene osnovne slojeve neuronskih mreža korištenih mreža te pregled dosadašnjeg rada. I za kraj, pokazujemo implementaciju generativnih suparničkih mreža temeljenih na progresivnom rastu i njihovu primjenu pri generiranju slika ljudskih lica koristeći skup podataka CelebA rezolucije 64 x 64 te vrednujemo rezultate.

Abstract (english)

Generative adversarial networks are a powerful class of deep learning models and a particularly active field of research. In this work, we present the basic idea of generative models and the concept of generative adversarial networks. Since this model is, in the broadest sense, based on the cooperation of two neural networks, we present a broad overview of this concept as well. We note the neural layers usually used in such a system and the work done so far. Finally, we implement a variant of this concept based on progressive growing and apply it to generation of images of human faces of 64 x 64 resolution using the CelebA dataset and evaluate our results.

• Keywords: generativne suparničke mreže; duboko učenje; generiranje slika; generativni model; progresivni rast mreža
• Keywords (english): generative adversarial networks; deep learning; image generation; generative models; progressive growing
• Language: croatian
• URN:NBN: urn:nbn:hr:168:191099
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 09:43:46