Interaktivno modeliranje terena z nevronskimi mrežami

diplomsko delo

Robi Novak (Author), Damjan Strnad (Mentor), Štefan Kohek (Co-mentor)

Abstract

V diplomskem delu smo implementirali generiranje terena z globokim učenjem. V ta namen smo uporabili pogojno generativno nasprotovalno mrežo (angl. conditional generative adversarial network) za tvorbo slike terena iz skic, kjer so označeni poteki vrhov in dolin. Naučena mreža omogoča interaktivno modeliranje terena, tako da uporabnik dopolnjuje skico, mreža pa sliko terena sproti prilagaja uporabnikovemu vnosu. Implementirali smo tudi prostorsko vizualizacijo generiranega terena z uporabo programskega vmesnika OpenGL.

Keywords

generiranje terena;modeliranje terena;nevronske mreže;diplomske naloge;

Data

Language:	Slovenian
Year of publishing:	2018
Typology:	2.11 - Undergraduate Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	R. Novak
UDC:	004.032.26(043.2)
COBISS:	21859606
Views:	869
Downloads:	193
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Interactive terrain modeling with neural networks
Secondary abstract:	In our work, we have implemented terrain generation using deep learning. We have applied conditional generative adversarial networks to terrain synthesis from ridge and valley sketches. A trained network allows for interactive terrain modeling where the user completes the sketch, while the network adjusts the terrain to the user's input. We also implemented a spatial visualization of generated terrain using OpenGL.
Secondary keywords:	terrain generation;terrain modeling;neural networks;
URN:	URN:SI:UM:
Type (COBISS):	Bachelor thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije
Pages:	VIII, 26 f.
ID:	10956687