|
|
3D object classification using neural networks
Krabec, Miroslav ; Křivánek, Jaroslav (advisor) ; Šikudová, Elena (referee)
3D Object Classification Using Neural Networks Bc. Miroslav Krabec Classification of 3D objects is of great interest in the field of artificial intelligence. There are numerous approaches using artificial neural networks to address this problem. They differ mainly in the representation of the 3D model used as input and the network architecture. The goal of this thesis is to explore and test these approaches on publicly available datasets and subject them to independent comparison, which has not so far appeared in the literature. We provide a unified framework allowing to convert the data from common 3D formats. We train and test ten different network on the ModelNet40 and ShapeNetCore datasets. All the networks performed reasonably well in our tests, but we were generally unable to achieve the accuracies reported in the original papers. We suspect this could be due to extensive, albeit unreported, hyperparameter tuning by the authors of the original papers, suggesting this issue would benefit from further research. 1
|
|
|
Procedural generation of cities in 3D
Krabec, Miroslav ; Gemrot, Jakub (advisor) ; Beneš, Jan (referee)
Title: Procedural generation of cities in 3D Author: Miroslav Krabec Department: Department of Software and Computer Science Education Supervisor: Mgr. Jakub Gemrot Abstract: During development of computer games great amount of time is spent on creating game environment. For that reason there is an effort to generate this environment procedurally. One of the interesting areas is generation of cities. Algorithm (Weber et al., 2009) is fully geometrical and offers considerable freedom in parametrization of the city, however there has not been an open implementation of this algorithm. Our work offers such an implementation and includes the possibility of 3D visualization in Unity 3D, a tool designed for development of computer games. Here we emphasize the ease of using custom 3D models of buildings, roads and crossroads. Resulting software can help game developers to generate cities for their games. However it can generate only relatively small cities (several thousands of buildings), but in context of computer games this is usually sufficient. Resulting city is very parameter sensitive and it is not trivial to achieve desired outcome. Further research would be needed for evaluation of generated cities. Keywords: procedural generation, cities, 3D
|
|
|
Testing the impact of quarterly results and analysts' expectations on prices of selected equities
Šoltés, Viktor ; Radová, Jarmila (advisor) ; Musílek, Petr (referee) ; Krabec, Miroslav (referee)
The aim of the article is to find the relationship between the growth and decline in the share price during the promulgation period of quarterly results of companies and surprise, either positive or negative in the quarterly results. Quarterly results are compared with the forecasts of analysts who publish their forecasts for quarterly results at Thomson Reuters. Relationship is confirmed statistically, where stock returns in the period is the dependent variable, independent variables are three -- return of the corresponding market index, excess impact -- measure of surprise in quarterly results in comparison with analysts' estimates and VIX index. Linear regression is used for testing of return and GARCH model is used for testing of volatility, there is focus on adaptation of actual volatility to the long-term average volatility after accidental shock.
|
guest ::
RSS feed.