|
|
Text Recognition in Natural Scenes
Tran, Tuan Hiep ; Surynek, Pavel (advisor) ; Holan, Tomáš (referee)
Title: Text Recognition in Natural Scenes Author: Tuan Hiep Tran Abstract: The main topic of our thesis is text recognition in natural sce- nes. This technology has finds many useful applications in the present world and has a great potential in the future one. We are analyzing the topic from theoretical but also from experimental perspective. Detailed analysis is done in theoretical part of this work. Based on this analysis, we propose inte- grating algoritm ER proposed by Matas and Neumann for letter detection with modified algorithm from Phan and co. for word recognition. Proposed method had been experimentally evaluated on testset ICDAR 2013 and ar- chieved recall 0.6 and precision 0.72. Our work also includes C++ library for text recognition, console and gui application for text recognition using the proposed method. 1
|
|
|
Social life of waste in the town of Police nad Metují
Tran, Tuan ; Grygar, Jakub (advisor) ; Kotyk, Lukáš (referee)
This bachelor thesis studies on the social life of waste in the town of Police nad Metují. The main theoretical source for the thesis is ''Social life of things'' by Arjun Apparudai, on which is based a big part of the thesis. The thesis examines the total trajectory of specific waste and tries to solve a question if waste can become a commodity. Special focus is given to the waste management and perception of waste from the perspectives of specific actors - citizens of Police nad Metují. The research was conducted in the spirit of qualitative methodology. The main sources of data are semi-structured interviews given by five participants, whose transcripts were analysed by content analysis. Additional data sources are observations within the region. Furthermore, a terrain probe in Vietnam metropolis, Ho Chi Minh City, whereas the data collected by this probe are given for comparison. The results of the thesis are showing that waste can become a commodity, although, just in specific cases. In the course of the research, infrastructure has emerged as an equally important topic, which has a major impact on waste management and its perception. Other major influencers identified by the analysis were ecology and the concept of usability.
|
|
|
The time-frequency relationship between spot and futures prices of crude oil
Tran Quang, Tuan ; Baruník, Jozef (advisor) ; Červinka, Michal (referee)
This thesis investigates the relationship between daily spot and futures prices for maturities of one, two, three and four months of West Texas Intermediate (WTI) crude oil. The data cover period January 1987-April 2015. Based on economic theory, the futures prices should be closely related to the spot price, which - in the case of crude oil market - this thesis analyses using wavelet-based approach. Main contributions of this thesis are findings in the field of time-frequency relationship of spot-futures prices of crude oil, where an alternative methodology - wavelet transformation - is used. The usage of this advanced method is also an additional contribution of this thesis because it allows us to rigorously study how co-movement (relationship) differs across frequencies/scales and time. In this thesis wavelet Coherence, wavelet bivariate correlation and relatively new method wavelet band spectral regression (WBLS) are used. This thesis brings 4 main findings. First, relationship between Futures and spot prices of crude oil is strong in all time-periods (frequencies/scales), which supports economic theory. Second and In contrary to the first finding, in the gasoline spot-futures market, we find that the relationship is strong mainly in higher scales (lower frequencies) while in lower scales (higher...
|
|
|
Text Recognition in Natural Scenes
Tran, Tuan Hiep ; Surynek, Pavel (advisor) ; Holan, Tomáš (referee)
Title: Text Recognition in Natural Scenes Author: Tuan Hiep Tran Abstract: The main topic of our thesis is text recognition in natural sce- nes. This technology has finds many useful applications in the present world and has a great potential in the future one. We are analyzing the topic from theoretical but also from experimental perspective. Detailed analysis is done in theoretical part of this work. Based on this analysis, we propose inte- grating algoritm ER proposed by Matas and Neumann for letter detection with modified algorithm from Phan and co. for word recognition. Proposed method had been experimentally evaluated on testset ICDAR 2013 and ar- chieved recall 0.6 and precision 0.72. Our work also includes C++ library for text recognition, console and gui application for text recognition using the proposed method. 1
|
|
|
Oligopoly Markets in Virtual Economy
Tran Quang, Tuan ; Skuhrovec, Jiří (advisor) ; Adam, Tomáš (referee)
This work will focus on the virtual World of Warcraft market where mainly I am going to focus my research on the behavior of monopolists and Oligopolists in this market and then trying to verify their pricing policy with known microeconomic models. This thesis finds that the behavior of monopolistic and oligopolistic markets in a virtual economy matches the prediction of standard microeconomic and that this harmony between theory and observation is, to some extent, even more clear than in the case of most real markets. Thanks to econometric analysis the empirical verification of the correctness of the economic game theory and possibly promoting the theory of regulation is another contribution of this work.
|
guest ::
RSS feed.