Automatsko bojanje crno-bijelih fotografija putem dubokog učenja

master's thesis

• Title (english): Automatic Colorization of Black and White Photos Using Deep Learning
• Author: Filip Anđel
• Mentor: Vladimir Čeperić (mentor)
• Committee member: Vladimir Čeperić (predsjednik povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Committee member: Tvrtko Mandić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-02-26, Croatia

Abstract

Predstavljen je problem kolorizacije slika te razvoj metoda kolorizacije počevši od ručnih metoda pa sve do tehnika dubokog učenja. Tehnike dubokog učenja detaljnije su opisane s posebnim naglaskom na transformerske modele, od kojih je jedan odabran, objašnjen i istreniran s ciljem koloriziranja crno-bijelih slika u stilu slikara Gustava Klimta. Opisan je proces treniranja, postignuti rezultati, ograničenja i moguća poboljšanja.

Abstract (english)

The problem of image colorization and the development of colorization methods, starting with manual methods and ending with deep learning techniques, are presented. Deep learning techniques are described in more detail with special emphasis on transformer models, one of which was selected, explained and trained with the aim of coloring black and white images in the style of the painter Gustav Klimt. The training process, achieved results, limitations and possible improvements are described.

• Keywords: kolorizacija; crno-bijele slike; duboko učenje; konvolucija; transformer; pozornost; Gustav Klimt
• Keywords (english): colorization; grayscale images; deep learning; convolution; transformer; attention; Gustav Klimt
• Language: croatian
• URN:NBN: urn:nbn:hr:168:057369
• 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: Open access
• Created on: 2024-10-12 19:58:24