prikaz prve stranice dokumenta Generiranje umjetnih slika lica pomoću generativnih suparničkih mreža
  Download
PDF 1.99 MB
master's thesis
Generiranje umjetnih slika lica pomoću generativnih suparničkih mreža
2023. urn:nbn:hr:200:323810
Ćaleta, Mislav
Josip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
Department of Computer Engineering and Automation
Chair of Automation and Robotics

Cite this document

Ćaleta, M. (2023). Generiranje umjetnih slika lica pomoću generativnih suparničkih mreža (Master's thesis). Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek. Retrieved from https://urn.nsk.hr/urn:nbn:hr:200:323810

Ćaleta, Mislav. "Generiranje umjetnih slika lica pomoću generativnih suparničkih mreža." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2023. https://urn.nsk.hr/urn:nbn:hr:200:323810

Ćaleta, Mislav. "Generiranje umjetnih slika lica pomoću generativnih suparničkih mreža." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2023. https://urn.nsk.hr/urn:nbn:hr:200:323810

Ćaleta, M. (2023). 'Generiranje umjetnih slika lica pomoću generativnih suparničkih mreža', Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, accessed 22 April 2024, https://urn.nsk.hr/urn:nbn:hr:200:323810

Ćaleta M. Generiranje umjetnih slika lica pomoću generativnih suparničkih mreža [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek; 2023 [cited 2024 April 22] Available at: https://urn.nsk.hr/urn:nbn:hr:200:323810

M. Ćaleta, "Generiranje umjetnih slika lica pomoću generativnih suparničkih mreža", Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, Osijek, 2023. Available at: https://urn.nsk.hr/urn:nbn:hr:200:323810

Please login to the repository to save this object to your list.
Metadata
TitleGeneriranje umjetnih slika lica pomoću generativnih suparničkih mreža
Title (english)Generating Synthetic Images of Human Faces by Generative Adversarial Networks
AuthorMislav Ćaleta
MentorRobert Cupec (mentor)
Committee memberRobert Cupec (predsjednik povjerenstva)
Committee memberRatko Grbić (član povjerenstva)
Committee memberEmmanuel Karlo Nyarko (član povjerenstva)
GranterJosip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
(Department of Computer Engineering and Automation)
(Chair of Automation and Robotics)
Osijek
Defense date and country2023-09-14, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Artificial Intelligence
Abstract
Glavni zadatak rada je bio napraviti model suparničkih generativnih mreža čija će generatorska mreža nakon učenja uspješno generirati slike ljudskih lica. Na početku je objašnjena sva potrebna teorija za shvaćanje modela koji su u radu napravljeni i korišteni. Nakon toga je prikazana potrebna priprema za sam proces izgradnje modela i njihovog učenja. Na kraju su svi modeli objašnjeni i pojašnjeni su razlozi za promjene koje su napravljene na modelima da bi se dobili bolji rezultati. Također su prikazani rezultati i objašnjeni su nedostatci rezultata i kako pristupiti poboljšanju. Poboljšanja koja su primijenjena u radu kako bi se postigli željeni rezultati proizlaze iz teorije strojnog učenja, točnije metoda iz dubokog učenja koje su se pokazale kao dobre metode za rješavanje problema koji se pojavljuju kroz rad. Svako poboljšanje je objašnjeno i naveden je jasan razlog zašto je napravljeno i kako bi teoretski trebalo utjecati na rezultate modela. Budući da su slike kroz verzije modela znatno bolje može se zaključiti da ovi postupci nisu učinkoviti samo u teoriji već su pomogli u generiranju kvalitetnijih slika.
Abstract (english)
The main task of this paper was to create a model of adversarial generative networks whose generator network after learning will successfully generate images of human faces. At the beginning, all the necessary theory for understanding the models that were created and used in the paper was explained. After that, the necessary preparation for the process of building the models and their learning is presented. At the end, all models are explained and the reasons for the changes made to the models for the purpose of getting the better results are clarified. The results are also presented. The shortcomings of the results and how to approach the improvements are explained. The improvements that have been applied in this paper to achieve the desired results stem from machine learning theory, more specifically deep learning methods that have proven to be good methods for solving problems that arise through this work. Each improvement is explained and a clear reason is given why it was made and how it should theoretically affect the results of the model. Because the images with different model versions are significantly better it can be concluded that these procedures are not efficient only in theory but also that they helped to generate higher quality images.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:200:323810
Study programmeTitle: Graduate University Study Programme in Computer Engineering
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2023-10-02 11:20:49