Bachelor theses
Algorithms for Semantic Analysis
Author
Jakub Švehla
Year
2014
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jiří Hanousek
Department
Automated text shortening
Author
Ivo Sklenář
Year
2015
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Karel Klouda, Ph.D.
Department
Summary
Goal of this thesis is to research and implement several algorithms for automatic text summarization. This thesis does a detail analysis of four algorithms and apart from text summarization algorithms, it also analyzes algorithms and methods used for natural language processing. At the end of this work, these algorithms were compared between each other for Czech and English languages and for each language, the best method of summarization was identified.
Detection of Phishing Messages
Author
Tomáš Duda
Year
2015
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Martin Kopp
Department
Summary
This bachelor thesis deals with the detection of phishing messages written in Czech or English language. Common features of phishing e-mails and present countermeasures are described. Based on gained information, an algorithm for phishing detection is proposed and implemented in Java. The algorithm is tested on messages written in Czech and English language.
Software application for project management using gamification elements
Author
Jan Slifka
Year
2015
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
The thesis explores gamification and gamification elements and mechanisms. It analyzes current use of gamification as a supportive tool for motivation, especially when working on serious tasks. The goal is to design and implement a new application that focuses on project management, task management, workers control and their motivation to work.
Design of a Module for Information Extraction from Documents
Author
Petra Krnáčová
Year
2015
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Department
Summary
The aim of this bachelor thesis is to design and to implement a module for Named Entity Recognition using rulebased approach. Another part of bachelor thesis describes existing approaches and tools used by Named Entity Recognition task, especially in Czech environment and in the extraction task in scanned documents transformed into text using Optical Character Recognition method.
Implementation of a Module for Sentiment Analysis
Author
Danny Dočekalů
Year
2015
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
This bachelor thesis deals with sentiment analysis of the Czech language. A recherche of the contemporary methods used in Czech was carried out. According to the recherche, a solution using machine learning was proposed and also a list of text attributes for that purpose. Java programming language was used to implement functions for text analysis and attribute detection. The quality of the attributes was tested using machine learning algorithms.
Web application for task management
Author
Jiří Ketner
Year
2016
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Zdeněk Rybola, Ph.D.
Department
Summary
The main objective of this Bachelor thesis is the implementation of web application for task management. Recherche of the term gamification and survey of the current applications on the market were made. Obtained information were utilized for the concept creation of own gamified application. It was implemented using freely available software tools after database design and deployment diagram creation. Function and user friendliness testing of the application were made by several users at the end. A part of the work is an installation guide.
Extraction of stamps from scanned documents and their preprocessing
Author
Jakub Šiller
Year
2015
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Daniel Vašata, Ph.D.
Department
Summary
This bachelor thesis focuses on identification and extraction of colour stamps from scanned documents. It describes segmentation of a document page into candidates, feature extraction, classification and post processing of extracted stapms like contrast improvement, noise reduction or rotation to basic position. The thesis also includes summarized previous approaches of stamps extraction, proposal of a solution and its implementation based on machine learning that uses k nearest neighbours algorithm. At the end of the thesis, the tests results are listed.
Forensic Analysis of Organized Groups using Data from Electronic Communication Protocols
Author
Ondřej Nový
Year
2015
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Martin Kopp
Department
Summary
This work tries to make forensic analysis of electronic data, particulary electronic communication, easier. Deals with problem of finding among communication participants the most important ones and with problem of finding tightly connected groups of them. The problems are solved using various algorithms for centrality measurement and community detection. Algorithms have been implemented in Java and tested on data used in real investigation of Enron corporation. Testing has shown that the algorithms are relatively successful, but none of them is significantly better than the others.
Identification of areas of common importance in a scanned document
Author
František Haifler
Year
2016
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Mgr. Jitka Rybníčková, Ph.D.
Department
Summary
This document discusses the problem of segments identification, which consist of data of similar meaning. First part analyses individual steps of such approaches including image preprocessing. Next part describes chosen method for solving this problem and its implementation. In conclusion are discussed achieved results and eventual adjustments for their improvement.
Identification of the functional style of the document
Author
Svetlana Ekimova
Year
2016
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
The thesis deals with the identification of the functional style of text documents. The identification task is solved as a classification task: the functional style of a document is detected by means of methods which are being used for the text classification. To learn the classifier, text documents are represented as a vector of attributes and labeled by a value describing their functional style. They are then given as input to a supervised machine learning method. The success rate of the learned classifier is evaluated by using k-fold cross-validation.
Detection of Glass Panes Defects
Author
Kristýna Tauchmanová
Year
2016
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
This bachelor's thesis deals with detection of glass defects. An algorithm able to detect and classify various types of defects is suggested based on literary research. The algorithm uses image processing methods, especially edge detection and image segmentation. It was tested on images of glass, which were shot by a line scan camera.
Detection and tracking of individuals on the basis of records from a camera system
Author
Petr Wudi
Year
2017
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Miroslav Balík, Ph.D.
Department
Summary
This thesis describes algorithm able to detect people in camera record and to track them. Problem is reduced to case of camera placed above people's heads. Object of the thesis is to compare various approaches and propose accurate real-time algorithm.
Detection of goods presence in customer's hand
Author
Oliver Keruľ-Kmec
Year
2016
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
This Bachelor Thesis deals with the detection of goods in the customer's hand displayed on the camera system. The record composed of the RGB frame and the depth map was produced by the Kinect device. The aim of this Thesis is to design, then to implement the algorithm that would be able to recognize whether the goods are taken out or put on the shelf. The theoretical part encompasses the analyzed algorithms for purposes of the attributes extracting and the basic techniques of the image processing. The Thesis covers the analysis of the proper and applicable algorithms for the classification, as well. Designed algorithm works with Gradient boosted trees classifier which classifies the frames according to 60 chosen attributes. By means of SIFT algorithm it is possible to find out in what direction the hand moves. Providing these two pieces of information, we managed to generate the sequences describing the hand's movement. Out of the sequence nine attributes were extracted which were used for classifier training. Onward, the classifier was able to decide whether the goods had been taken or put on the shelf. The accuracy of this approach succeeded in reaching 71%. The algorithm was implemented in the C++ language and employed the OpenCV library, as well.
Gaming environment with artificial intelligence for motivational web application
Author
Jakub Kulík
Year
2016
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
Bachelors thesis describes an extension of a motivational web based application and its strategy game by a computer opponent with intelligent behavior. The application includes rule-based system for control of opponents and multi-agent system to simulate the movement of armies and interaction between them. The thesis also includes analysis and implementation of a random island generator. Because of the target group the application is not time consuming and it doesn't unnecessarily punishes inactive team members. The artifical inteligence is curtailed or completely inactive if necessary.
Preprocessing of Images of Tire-thread Patterns
Author
Filip Pavlas
Year
2016
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
This bachelor thesis deals with creating an image of the tire tread. The resulting image is obtained based on the number of partial images of a rotating tire. The partial images were taken by a matrix camera. Methods for obtaining images, identification tires, geometric correction and mosaic creating are described. Specific solutions is designed. Solution is implemented in C# using OpenCV and emguCV libraries.
Automatic reconstruction of missing data
Author
David Záleský
Year
2016
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
prof. Dr. Ing. Petr Kroha, CSc.
Department
Summary
This paper focuses on errors in measuring customers' activity in a supermarket. It describes the procedure of gathering data, identifies potential errors, their causes, and potential solutions. The solutions are then analyzed to determine their usability with provided data. Selected solutions are implemented in programming language Python.
Web application for designing a model of railroad trackage
Author
Martina Egrtová
Year
2017
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Radek Dobiáš, Ph.D., MBA
Department
Summary
The subject of this thesis is the implementation of a web application that allows an interactive graphic design of the model railroads. The main purpose is to simplify work on designing model railway tracks and to provide to user an overview. The web application is intended for users interested in model railways. The thesis describes the technologies used to create the application, available solutions, analysis and design of the application, its implementation and testing.
Detection and tracking of persons in camera video records
Author
Sultan Ongarbayev
Year
2017
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Josef Gattermayer, Ph.D.
Department
Summary
The main goal of thesis is to propose and implement software application that will be able to detect and track people using several video cemaras. User of this application will be able to record videos from cameras, process them and get easy-to-read output. The thesis explores existing technniques, describes selected method and presents implementation details. Results and testing approach are also presented.
Methods of Artificial Intelligence for Motivational Gaming Environment
Author
Matěj Sochor
Year
2017
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
This bachelor thesis examines possibilities of an artificial intelligence for games
to adapt its strategy to its opponent in a complex strategy game, which
forms part of the motivational task management application Taskino.
Such an adaptation would improve the gameplay. It describes and evaluates various methods of adaptive artifical intelligence.
From these methods, a dynamic scripting method was chosen for the practical part.
Dynamic scripting was adjusted for the game and was implemented in it.
Its quality was assessed according to the behaviour of the adaptive artificial intelligence (created by dynamic scripting)
and its performance in the matches against three static artifical intelligences representing
offensive, defensive and constructive game strategies. The thesis draws conclusions about
the game mechanics of the game from the course and results of these matches
and recommends enhancements to improve its gameplay.
Analysis and description of a tire tread
Author
Radka Vopatová
Year
2018
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Ivan Šimeček, Ph.D.
Department
Summary
This bachelor thesis deals with the analysis and description of tire tread imprint from photograph. The goal is to explore recent methods which are used for description of tire tread pattern and design own method for its representation. Selected representation will be implemented in MATLAB environment and then it will be tested on real data.
Design of algorithms to support writing reviews
Author
Jakub Vašíček
Year
2017
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
The goal of this thesis is to design, implement and test algorithms which will assist the user in writing a review of a final thesis. The assistance consists mostly of recommending formulations that the user can integrate into their text. Relevant formulations are identified according to the evaluation given by the user. The formulations are also split into groups by their meaning so that it is possible to interchange formulations with similar meanings. I implement the algorithms in the form of a web application. This work can be of use to reviewers and supervisors of student's theses --- the reviews often contain repeating formulations which are suggested by the application and the user can concentrate solely on evaluating the thesis.
Detection of products in the customer's hand by analyzing images from a thermocamera
Author
Lukáš Brchl
Year
2017
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Jan Janoušek, Ph.D.
Department
Summary
This thesis deals with the detection of goods in the customer's hand by analyzing images from infrared camera. The camera is located above the shelf, monitoring behavior of the customer standing in front of it. Acquired data contains informations about the temperature of the objects in the view. Using these temperature data, it is possible to segment hand and examine the presence of the goods on the scene. The aim of the thesis is to explore possible solutions of this task, to propose a solution and subsequently implement it. There are several approaches to the problem and several possible solutions are proposed. The FLIR A65 thermal camera with the eBUS SDK tool is used to capture data. The algorithms are implemented in Java using the OpenCV library. The evaluated background subtraction algorithm achieves an overall classification accuracy of 88 %.
Optimizing of methods for data preprocessing for the best possible classification
Author
Jan Pancíř
Year
2018
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
The thesis deals with the efficiency of the process of extracting knowledge from the data. It focuses on optimizing the order of pre-processing methods and their parameters for the specific classifier. An algorithm has been designed to create a plan to pre-process input data so that the learned classifier achieves the highest accuracy of classification. The genetic algorithm was used to optimize and find the plan. A Java application implementing this algorithm has been developed to find a pre-processing plan for any numerical dataset using the following data preprocessing methods: discretization, normalization, dimensionality reduction, remote values removal, class balancing and missing
values imputation. The results were tested on several real datasets. The algorithm improves the classification accuracy by an average of 4-9 %. This is a tool that allows you to fully automate the pre-processing process. Eventually it can be used as a help tool for a knowledge expert to create a pre-processing plan.
Translator of flowcharts from Java to C/C++
Author
Marek Tornóci
Year
2018
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
This work deals with a design and implementation of the application, which will be able to translate chosen already implemented blocks in the Java programming language to the C++ programming language. Those methods represent various image processing operations which can a user freely combine in the interactive environment called Surmon. The result of the thesis is implemented application in the Java programming language integrated in the Surmon environment.
Evaluating the quality of the geometric shapes
Author
Daniel Němec
Year
2018
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jiří Chludil
Department
Summary
This thesis deals with the design of an algorithm for evaluation of the quality of hand-drawn geometric shapes using known metrics or metrics from own design. The OpenCV image processing library is used to design and implement the algorithm, and quality metrics for a given geometric shape are used for quality evaluation. The proposed solution is tested on real data.
The advantage of the application is the ability to effectively compare each result with one another, filtering between them and displaying the development over time. The application allows the user to improve in drawing geometric shapes. I have found out the differences in the quality of the line depending on its inclin in degrees.
Application for eBay auction monitoring and management
Author
Tomáš Polačok
Year
2019
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Vojtěch Jirkovský
Department
Summary
This bachelor thesis describes creation of software application for working with
eBay auction portal. It includes analysis of the current state of searching and
monitoring items on a given portal, as well as analysis of the requirements for
the final application. Consequently, it contains a design of my own solution
created on the basis of this analysis and the related selection of suitable technologies. Lastly it describes implementation and evaluation of the results of
the created application.
Design of a camera system for scanning a tire tread
Author
Daniel Bohuněk
Year
2022
Type
Bachelor thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Tomáš Kalvoda, Ph.D.
Department
Summary
The objective of this bachelor's thesis is to design a camera system for scanning tire treads of passing vehicles in a standardized form.
After researching existing solutions for scanning tire treads a new camera system is proposed, which captures and unwraps the tire tread of vehicles stopped at a traffic light. A convolutional neural network is used to find the location of the tire in an image.
The camera system was tested in real traffic at multiple traffic lights. 78 % of vehicle tires are captured correctly. The implemented algorithms correctly process the tire tread 89 % of time.
This new camera system can be used to collect large amount of tire tread samples, which might otherwise be impractical and expensive to obtain. This data is useful for example in forensic analysis, to pair a tire tread with tracks left on the ground.
Master theses
Utilization of Smartphone for Measurement of Patient Walking Characteristics
Author
Jan Sedláček
Year
2014
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Tomáš Borovička
Department
Detection of links based on the phonetic similarity of words
Author
Václav Tarantík
Year
2015
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Mgr. Jan Starý, Ph.D.
Department
Summary
The work deals with the study of algorithms comparing words based on its phonetic similarities and with the implementation of these algoritms in the SVAT environment. The part of this work is also the design of new algorithm based on Metaphone algorithm and focused mainly on the typical characteristics of the Czech language and its further comparison with existing algorithms. The output of the work is an application written in the Java SE programming language usable as Netbeans IDE plugin and also the description of measurement results for implemented methods including graphs.
Motivational application for project administration using gamification elements
Author
David Kulík
Year
2015
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Michal Valenta, Ph.D.
Department
Motivational web application using gamification elements
Author
Michal Kvasnička
Year
2015
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
This master thesis describes the process of design and implementation of motivational web application for task management. The main goal of the application is to be useful for organizing team work. The application extends project management by using gamification elements, thereby it motivates users to be more hardworking. The application aims to be fast and intuitive, therefore it was implemented with use of modern web technologies. The thesis also covers different types of project management and describes the most widely used elements of gamification.
Automatic text summarization
Author
Šimon Hlaváč
Year
2015
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
This work presents the basic methods used in automatic text summarization and genetic algorithms. Furthermore, system of automatic summarization based on graph structures and Markov chains was designed, implemented and properly tested. This study also discusses learning of proper setting of importance weights of individual methods used in summarization by naive approach and genetic algorithms, which were also implemented and properly tested. System also includes possibility of parallel processing and use of caching to speed up its process.
Neural networks with memory
Author
Ondřej Kužela
Year
2016
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Josef Pavlíček, Ph.D.
Department
Summary
Neural networks with memory are the family of the neural networks that except the classic memory for the long-term dependencies, in a form of the weights, also contain another form of a memory. Such a memory serves to retain the mid-term, sometimes also called long-short-term, dependencies and can be of two different types, either internal or external. Within this thesis I offer a summarizing overview of the family of the neural networks with memory. Based on the analysis of the existing models I also propose a new model of the Recurrent Neural Modules with External Memory. This model offers a new and innovative approach to the usage of the external memory within the neural networks, since it deploys the external memory on the scope of parts of the network and thus deploys multiple external memories within one network. The performance of the newly proposed model was evaluated on the Air Travel Information System (ATIS) dataset.
Information extraction about persons in images from a camera
Author
Vojtěch Haur
Year
2016
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
Image data of customers in a supermarket contain valuable information, which can be used by the supermarket management for business decisions. The task of this work is to deal with raw dataset of such images and explore the possibilities of automatized information extraction. The approach is to prepare an annotation of the dataset, research the methods used for similar tasks, implement them and evaluate how they can perform on this data. The aim is to open the topic for future works and provide them with a basis upon which they can expand.
Extraction of stamps from scanned documents and their preprocessing
Author
Peter Šourek
Year
2016
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Radek Richtr, Ph.D.
Department
Summary
The aim of this thesis is to design, implement and test algorithm for stamp
detection and extraction from scanned documents. Proposed algorithm is using
machine learning algorithms for stamp classification using feature set, which is
also designed in this thesis. Finally, testing of developed algorithm is done on
data set from server justice.cz and best-performing machine learning algorithm
is chosen based on test results.
Detection and tracking of individuals on the basis of records from a camera system
Author
Michal Řapek
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Miroslav Balík, Ph.D.
Department
Summary
The task of this work is to implement a library of functions for people detection and tracking. For this purpose the thesis contains a summary of the most commonly used methods in similar cases, ie for detection of persons in depth map when viewed from above.
The second part contains a description of implementation and results of testing.
Graphics user interface for designing a visualization element
Author
David Klingenberg
Year
2016
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Pavel Štěpán
Department
Summary
Goal of this thesis is to develop a tool for creating a plan of the mall and visualisation
application for visualizing heatmap on the created plan. Visualization
application will display heatmap based on the input data. Both applications
are programmed in JavaScript and run in the web browser environment. Next
goal is to integrate visualization application into reporting tool Sureport.
Optimization of Optical Biosensor by Evolutionary Methods
Author
Jan Horký
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Ivan Šimeček, Ph.D.
Department
Summary
This thesis deals with optical component of biosensor. It explains how various
attributes of this component change its ability to concentrate scattered beams
into optical fibre. Then it tries to find the best possible combination of these
attributes with the help of genetic algorithm.
Feature selection methods for classification and regression problems
Author
Pavel Verner
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
Master's thesis deals with analysis of feature selection methods and its comparison. In
the first part of the work we will divide method using their properties and common signs.
Further we will analyse its mathematical basis and the principle of functioning. Practical
part will focus on comparison selected methods using multiple data samples. Methods are
compared based on their performance, time complexity and relationship to datasets. At the
end of each section there is a recommendation, which reflect observed measurements.
Detection of a license plate position from camera records of moving cars
Author
Vladislav Jásek
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Jan Janoušek, Ph.D.
Department
Summary
This thesis deals with the problem of license plates detection in moving scene. Some of the current approaches to the problem are summarized in this thesis. Several state-of-the-art techniques of tracking and detection are introduced and detection framework solving the problem is proposed. Practical implementation is described and experimentally evaluated.
Interactive web interface to design flowcharts
Author
Jan Tesařík
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jakub Novák
Department
Summary
The aim of this thesis is to design and implement appropriate web interface or in other words, a front-end of a web application, for a back-end that is currently being developed. The project as a whole is called Surmon and it allows users to create algorithms by constructing block schemes using a graphical interface. Block schemes are generally diagrams depicting a flow of data through the blocks of the scheme, on which the data is processed and used for various calculations. Some blocks serve as the input and some as the output of the scheme. In this thesis, I will be dealing with the design and implementation of the creation of these schemes on the client's side, communication with the back-end, but the functionality of the blocks is the back-end's responsibility. The functionality can be for example image processing or mathematical functions among many others.
Application to support writing scientific papers
Author
David Vondraš
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Jiří Novák, Ph.D.
Department
Summary
The aim of this diploma thesis is to design a web application that will facilitate the creation of scientific articles by offering pre-prepared sentences and text blocks depending on the chosen topic. The application contains the features needed for whispering sentences and for creating a complete article, including inserting attachments and generating quotations. The work includes the description of application design, implementation and testing.
Detection and tracking of persons in camera video records
Author
Jan Kozák
Year
2018
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Karel Klouda, Ph.D.
Department
Summary
The analytical part of this thesis offers an introduction to the field of person detection and tracking. The most commonly used methods utilizing single and multiple static cameras are categorized and described. The proposed tracking algorithm starts with a multi camera detection approach that uses the geometric properties of the scene to limit the effects of partial and complete occlusions. Hungarian algorithm is used to assign detections to the track positions estimated by the Kalman filter.
Analysis of bank data
Author
Evgeniya Nenenko
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Department
Summary
The aim of this work was to analyse bank transactional data in order to find patterns in the customers' behaviour. The main focus being periodic pattern mining in time series databases. To provide a suitable analytical framework, the research of the existing solution was made with a later adaptation of available algorithms to suit the requirements of the assignment. Finally, the result evaluation took place with suggestions regarding framework settings and future usage.
Analysis and prediction of user behavior on auction web sites
Author
Lukáš Kratochvíl
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
The goal of this thesis was to analyze behavior of users on auction websites and to predict final prices of auctions. Used data was downloaded from Aukro.cz website by using TextMiningSuite util. Design of preprocessing, sets of atributes (which meet required level of dependency according to final price) and learning algorithms was followed by implementation realizing mentioned procedure. At the end of the thesis a set of the most user influential atributes was defined. Then the best auction final price predicting algorithm (and its settings) was found. This section was followed by suggestion of steps that may lead to improve results of predictions.
Automatic naming of groups of words
Author
Jan Effenberger
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Petr Špaček, Ph.D.
Department
Summary
This work solves the development of application for automatic cluster labeling, using external databases of knowledge. In my thesis I analyze the possibilities of description and commonly used methods for the cluster labeling. I used the practical experience with the focus on further expandability and development. As an external database, I used WordNet and Microsoft Concept Graph at work, and I used Java as a programming language. I managed to create an application that makes it easy to change the labeling algorithm and other parts, which is the main benefit of my work, and I also put together two labeling algorithms -- each for one knowledge base. The results of the labeling in the functional testing proved to be adequate and satisfactory with respect to the quality of the input data.
Applications supporting teaching of mathematics at elementary schools
Author
Petr Zavadil
Year
2017
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Karel Klouda, Ph.D.
Department
Summary
This diploma thesis contains design of tablet application with augmented reality. The application is designed to be controlled by manipulation with real objects in surroundings of a tablet device. The major part of thesis deals with image processing task of recognition and classification of defined real objects. The output of thesis is working sequence of image processing algorithms, which is able to recognize given objects in image. The software part of thesis is implemented in environment of Matlab. The part which is aimed to image processing is developed according to CRISP-DM methodology.
The use of artificial intelligence for analysis of escape from a burning room
Author
Jiří Kukačka
Year
2018
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Tomáš Kalvoda, Ph.D.
Department
Summary
Aim of this thesis is to explore possibilities of using artificial intelligence algorithms and machine learning in field of fire protection related optimalization of buildings, and set basics for automated design checking of buildings.
Detection of material defects on foamed insulating panels
Author
Tomáš Duda
Year
2018
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Ivan Šimeček, Ph.D.
Department
Summary
This master's thesis deals with the automatic detection of material defects on foamed insulating panels using methods of image processing. The process of foamed glass production and current approach to output quality control by a human worker is described. The description of installed hardware for image data acquisition is provided. Related systems for automatic material inspection are reviewed and an analysis of various methods for image texture description is provided. Conceptual design of foamed glass panels inspection system is presented. Acquired images of panels are described and an annotation application is developed. A suitable image preprocessing algorithm is proposed as well as methods for detection of different kinds of foamed glass defects. The final design of detection algorithm is supported by measurement of the accuracy of several methods. The proposed algorithm is implemented and the final accuracy of the inspection system is measured. The results are discussed and possible future improvements are proposed. The developed system was successfully deployed in a production environment.
Quality Measuring of Quantitative Resources for Automated Forecasting of New Technologies
Author
Michal Režnický
Year
2018
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Josef Gattermayer, Ph.D.
Department
Summary
This master thesis is dealing with measurement of quality of quantitative sources for automated forecasting of emerging technologies. Analytical part of this thesis covers different forecasting fields from the technical point of view as well as economical. Fundamental part of this thesis is design of original metrics for measuring the quality of quantitative sources for automated forecasting of emerging technologies. The basis for designed metrics lays in the analysis of methods of measuring sources quality in other fields. The sources are rated by the rules of designed metrics. The metrics are given weights byimportance and several sources quality are calculated.
Identification of the human emotional states based on a sequence of images of his face
Author
Martin Endršt
Year
2018
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
doc. Ing. Mgr. Petr Klán, CSc.
Department
Summary
Emotion expression is an important aspect of human to human communication. Recognizing the emotional state of a person can help us better understand complex rethorical devices such as irony, understand the gravity of described situation and infer other information that is often not expressed as part of the verbal communication channel. With the growing popularity of integrated human-machine interfaces automatic emotion detection has a great potential to improve the way we interact with machines. Since camera sensors are being integrated into almost all devices, emotion recognition based on facial expression is one of the viable methods for widespread use. Several models performing emotion recognition based on sequence of frontal facial images were proposed and implemented in this thesis. Because emotion is a dynamic psychical state, three different types of temporal context information for recognition were examined and compared. To ensure usability with real-time streams a wrapper framework consuming one frame at the time is proposed. Both deep-learning based and conventional types of classifiers were implemented. The best performing model achieved accuracy of 95.1% on the CK+ dataset.
People detection, tracking and biometric data extraction using a single camera for retail usage
Author
Lukáš Brchl
Year
2019
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Mgr. Jan Starý, Ph.D.
Department
Summary
This thesis proposes a framework that analyzes video sequences from a single RGB camera by extracting useful soft-biometric data about tracked people. The aim is to focus on data that could be utilized in a retail environment. The designed framework can be broken down into the smaller components, i.e., people detector, people tracker, and soft-biometrics extractor. The people detector employs various deep learning architectures that estimate bounding boxes of individuals. The tracking solution follows the well-known online tracking-by-detection approach, while the proposed solution is built to be robust regarding the crowded scenes by incorporating appearance and state features in the matching phase. Apart from calculating appearance descriptors only for matching, we extract additional information of each person in the form of age, gender, emotion, height, and trajectory when possible. The whole framework is validated against the dataset which was created for this propose.
Knowledge system to support BI-ZNS education
Author
Jan Horáček
Year
2019
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Tomáš Nováček
Department
Summary
The objective of this thesis was to create a new framework for teaching
knowledge-based systems in the subject Knowledge-based Systems (BI-ZNS)
at the Faculty of Information Technology.
In the first phase was done an analysis of knowledge-based systems. The
analysis was focused on defining the requirements and constraints of the target
problem domain that users will use for creating the knowledge-based system.
Based on the analysis was developed a basic concept of the tower defense game.
The game will serve as a problem domain for knowledge-based systems.
Subsequently, a detailed design of all game principles that will occur in
the game was made. Based on the principles and requirements for the entire
system, a modular framework and its components were designed.
Based on the analysis and design, the framework in Python was developed.
The framework contains a standalone playable computer game. Also,
the framework contains auxiliary modules and a basic skeleton for building
knowledge-based systems.
In the end, the thesis deals with the possibility of extending the framework
with additional functionality. Also, there is the possibility to create a new game
with re-using knowledge-based system modules. To enable an easy extension
and trouble-free use of the framework in teaching, the last part of the theses is
focused on testing of all framework functionality.
Web application for modular model railway design
Author
Martina Egrtová
Year
2019
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
Ing. Petr Špaček, Ph.D.
Department
Summary
The subject of this thesis is to extend current web application, which is based on my bachelor's thesis. This application allows interactive graphic design of track layout within the modules for model railway. The extension consists in the implementation of a complex design of the railway from the already designed modules and the possibility of designing a two-level railway. Another extension is improvement to the user interface and implementation of other features which aids modellers to design railways and makes it possible share it with community. Thesis describes analysis and design of extension functions, implementation of essential parts of the application and user testing.
Optimization of the painting placement problem using evolutionary techniques
Author
Martin Šafránek
Year
2023
Type
Master thesis
Supervisor
doc. RNDr. Ing. Marcel Jiřina, Ph.D.
Reviewers
prof. RNDr. Pavel Surynek, Ph.D.
Department
Summary
The thesis proposes a genetic algorithm with a novel chromosome representation and novel crossover implementation for solving the NP-hard problem of placing rectangular paintings on a two-dimensional grid. A chromosome is represented as multiple stochastic vectors (vector that contains non-negative elements that add up to one). Crossover is implemented as vector addition followed by normalization back to the stochastic vector. The proposed solution is tested on a generated dataset.