Mgr. Martin Mareš

Mgr. Martin Mareš

Ing. Ivo Petr, Ph.D.

Department of Applied Mathematics

The main goal of this thesis is to analyze published official notice boards in form of open data and present the results on the web. In order to analyze the compliance of the published official notice boards, a list of municipalities is obtained. In the next step, a list of links to all published official notice boards is obtained. These boards then get downloaded and have data extracted from them. If some of these boards have links to attached documents, those will get downloaded as well and text will be extracted from them. All of these processed data will be stored in a database. This data can be used later on. For example on the created website, where a user can look through the process data, or at one of many charts, showing statistics of the data.

Thesis on DSpace

Mgr. Martin Mareš

PhDr. Ing. Tomáš Evan, Ph.D.

Department of Software Engineering

Transparent bank accounts are one of the tools that support transparency in financing. Political entities are legally obligated to use transparent accounts. Such accounts typically contain a large number of transactions, and therefore, their detailed analysis can provide interesting information. The existence of a tool that would simplify public oversight of transparent accounts would benefit not only journalists but also other interested parties. This thesis focuses on the development of a web application that processes data from transparent accounts on bank websites and provides these data to public using an API. The application subsequently analyzes the obtained data, including statistical information, aggregation of transactions according to counterparties with a search for counterparties in transactions from other transparent accounts, and visualization of transactions over time. The application currently supports transparent accounts from three banks with a simple extension possibility for others.

Thesis on DSpace

Mgr. Martin Mareš

Ing. Jan Matoušek

Department of Software Engineering

Public procurements are one of the key tools for transparent and efficient management of public finances. The aim of this thesis is to create a web application that obtains data on public procurements in the Czech Republic and subsequently visualizes them using a set of interactive maps. The resulting application periodically retrieves data from the National electronic tool using web scraping. The application uses the process of geocoding to enrich the acquired data with the geographical location of each legal entity. The obtained and enriched data are visualized on maps, and the application supports filtering the displayed data by place of performance, contracting authority, and participant. Displaying public procurements data on a map helps to identify Czech as well as foreign locations where public funds from the Czech Republic go, which can help gain a broader context about the competing parties.

Thesis on DSpace

Mgr. Martin Mareš

doc. Ing. Ivan Šimeček, Ph.D.

Department of Applied Mathematics

This diploma thesis focuses on the automatic lip reading for Czech language. This automatic speech recognition is performed in uncontrolled environments that are characterized by varied lighting conditions, diverse backgrounds and multiple different speakers all of which complicate the visual processing of lip movements. This thesis inspects existing approaches for other languages and applies and evaluates similar principles to the Czech language using deep learning. Recognizing the limited resources in non-English datasets for lip reading, this work also involves the creation of a tailored train, validation and test dataset to facilitate model training and evaluation. The evaluation is done using newly created Czech lipreading testing dataset created with help of Czech Television as a part of this thesis.

Thesis on DSpace