Metoda polunadziranog učenja za procjenu više osvjetljenja u sceni

master's thesis

• Title (english): Semi-supervised Learning Method for Estimation of Multiple Scene Illuminants
• Author: Matej Salković
• Mentor: Sven Lončarić (mentor)
• Committee member: Sven Lončarić (predsjednik povjerenstva)
• Committee member: Marko Subašić (član povjerenstva)
• Committee member: Tomislav Petković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Automatska procjena boje osvjetljenja scene je složen računalni problem. Važna je jer mnoge metode za klasifikaciju i segmentaciju jako ovise o bojama u slici. Problem dodatno otežavaju situacije u kojima u sceni postoji više od jedne boje osvjetljenja. Ovaj diplomski rad bavi se metodom polunadziranog učenja za procjenu dvije boje osvjetljenja, tj. boju osvjetljenja u osvijetljenom dijelu scene te boju osvjetljenja u sjeni. Implementirano je i testirano pet međusobno različitih arhitektura koje se baziraju na kombinaciji dviju metoda za procjenu jedne boje osvjetljenja. Prva metoda koristi nadzirano učenje, dok druga metoda koristi nenadzirano učenje. Stoga se metoda za procjenu dva osvjetljenja smatra metodom polunadziranog učenja.

Abstract (english)

Automatic illumination color estimation of a scene is a complex computational problem. It is important because many methods for classification and segmentation heavily depend on the colors in the image. The problem is further aggravated by situations in which there is more than one illumination color in the scene. This master thesis deals with the method of semi-supervised learning for the estimation of two illumination colors, i.e., the illumination color in the illuminated part of the scene and the illumination color in the shadow. Five mutually different architectures based on a combination of two methods for estimating a single lighting color were implemented and tested. The first method uses supervised learning, while the second method uses unsupervised learning. Therefore, the method for estimating two illuminations is considered to be a semi-supervised learning method.

• Keywords: procjena boje više osvjetljenja; polunadzirano duboko učenje
• Keywords (english): multiple illuminants estimation; semi-supervised deap learning
• Language: croatian
• URN:NBN: urn:nbn:hr:168:392784
• 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: 2023-01-14 09:53:21