Modeli za generiranje trodimenzionalnih objekata

master's thesis

• Title (english): Models for Generation of Three-dimensional Objects
• Author: Bruno Sačarić
• 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: 2021-07-09, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Kod složenih modela dubokog učenja pojavljuje se problem nedostatka dovoljnog broja primjera za učenje. Moguće je poboljšati svojstva tih modela tako da se generiraju sintetski primjeri pomoću složenih modela generativnih suparničkih mreža. Ti modeli se brzo razvijaju i postaju sve sposobniji generirati realistične slike i modele. 3DGAN arhitektura omogućuje generiranje 3D modela veće preciznosti. U ovom radu je pokazana primjena te mreže na generiranje ultrazvučnih slika pukotina u materijalu, kako bi se povećanjem skupa primjera poboljšala generalizacijska svojstva mreža za klasifikaciju tih pukotina.

Abstract (english)

Complex deep learning models sometimes face the problem of not having enough learning examples. By generating synthetic examples with generative adversarial networks, it is possible to improve the capabilities of those networks. Those models are rapidly evolving and becoming more and more capable of generating realistic images and models. 3DGAN architecture makes it possible to generate 3D models of greater accuracy. This thesis shows the application of that network for generating ultrasound images of defects in materials, so that by enlarging the training set, we could improve the network's generalisation capabilities for classification of defects.

• Keywords: generativni suparnički modeli; neuronske mreže; umjetna inteligencija; strojno učenje; detekcija pukotina
• Keywords (english): Generative Adversarial Nets; Neural Networks; artificial intelligence; machine learning; detection of defects
• Language: croatian
• URN:NBN: urn:nbn:hr:168:462720
• 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 13:45:01