|
|
Anticurtaining - Image Filter for Electron Microscopy
Dvořák, Martin ; Dobeš, Petr (referee) ; Zemčík, Pavel (advisor)
Tomographic analysis produces 3D images of examined material in nanoscale by focus ion beam (FIB). This thesis presents new approach to elimination of the curtain effect by machine learning method. Convolution neuron network is proposed for elimination of damaged imagine by the supervised learning technique. Designed network deals with features of damaged image, which are caused by wavelet transformation. The outcome is visually clear image. This thesis also designs creation of synthetic data set for training the neuron network which are created by simulating physical process of the creation of the real image. The simulation is made of creation of examined material by milling which is done by FIB and by process displaying of the surface by electron microscope (SEM). This newly created approach works precisely with real images. The qualitative evaluation of results is done by amateurs and experts of this problematic. It is done by anonymously comparing this solution to another method of eliminating curtaining effect. Solution presents new and promising approach to elimination of curtaining effect and contributes to a better procedure of dealing with images which are created during material analysis.
|
|
|
In-situ Study of Surface Topography Changes in Elastohydrodynamic Contact
Šperka, Petr ; Dzimko, Marián (referee) ; Návrat, Tomáš (referee) ; Dobeš, Petr (referee) ; Hartl, Martin (advisor)
This dissertation thesis deals with elastohydrodynamic lubrication (EHL) of surfaces with real roughness. Systematic study of surface harmonic components behaviour, during their passage of the EHL contact, by other authors led to retrieval of a uniform principle according which individual harmonic components deform inside the contact. Although the amplitude attenuation model can provide rapid tool for more accurate surface assessment in design process it has not been validated on real surfaces till now. The thesis represents pioneering work on quantitative analysis of real surfaces EHD measurements. It contains original and new results with real and model roughness features that in many cases confirm theoretical expectations. Deeper understanding of surface roughness behaviour inside EHL contact, which is typical for machine parts like bearings, gears, cams etc., can help to improve design process of the components and ultimately enable machines and equipment to work with lower energy requirements and higher durability.
|
|
|
THIN LUBRICANT FILMS STUDY USING SPECTROSCOPIC REFLECTOMETRY
Čudek, Vladimír ; Dzimko, Marián (referee) ; Navrátil, Karel (referee) ; Dobeš, Petr (referee) ; Křupka, Ivan (advisor)
ectroscopic reflectometry is measurement technique that enables the study of the thickness and refractive index of thin layers. This thesis deals with its use for the study of lubrication films within EHD contact and verifies a new approach suggested in recent publications. It is focused on the development of an optical arrangement and mathematical model for the lubrication film thickness measurement within the entire EHD contact. This technique brings the possibility to study the pressure distribution within the contact area through the evaluation of changes in lubricant refractive index.
|
|
|
New Generation of Open-Source Tool Stranger Strings
Kňazík, Matej ; Dobeš, Petr (referee) ; Herout, Adam (advisor)
Cieľom tejto bakalárskej práce je opísať vývoj webovej aplikácie s názvom Stranger Strings. Projekt si ponechal svoje meno po internom firemnom nástroji, avšak vo všetkom ostatnom prešiel kompletnými zmenami. Účel Stranger Strings sa pozmenil, kód bol od základu prepísaný a celý projekt bol zmenený na open-source. Rovnako bolo redizajnované aj užívateľské prostredie za účelom rozšírenia cieľovej skupiny užívateľov. Táto práca podrobne popisuje tento proces a tiež vysvetľuje, akú úlohu zohralo užívateľské testovanie pri identifikácii hlavných problémov aplikácie. Výsledkom tejto práce je open-source nástroj na analýzu lokalizácie, ktorý pomáha užívateľom zlepšovať jej kvalitu.
|
|
|
Processing Sensor Data from a Wearable Device by Machine Learning
Hlavačka, Martin ; Dobeš, Petr (referee) ; Herout, Adam (advisor)
The goal of this master's thesis is to analyze the situation of wearable devices with the Android Wear operating system and recognition capabilities of various movement activities using neural networks. The primary focus is therefore on identifying and describing the most appropriate tool for recognizing dynamic movements using machine learning methods based on data obtained from this type of devices. The practical part of the thesis then comments on the implementation of a stand-alone Android Wear application capable of recording and formatting data from sensors, training the neural network in a designed external desktop tool, and then reusing trained neural network for motion recognition directly on the device.
|
|
|
Generative Neural Networks for Handwritten Text
Ševčík, Pavel ; Dobeš, Petr (referee) ; Hradiš, Michal (advisor)
The aim of this study was to create a generative neural network for handwritten text lines. The model produces variable-sized images of handwritten text lines based on the expected style. The proposed method exceeds existing models in the image quality and can be used to generate both individual words and entire lines of handwritten text. It combines the use of the attention mechanism to extract the features for each character from the text query and their arranging on the line by inserting spaces between them. The new approach allows more granular control of the symbol positions on the line, which leads to smoother style interpolations. In contrast to the previous approach, the proposed method uses the Gaussian filter to spread the individual symbols features to the surrounding area. This approach also allows to train the model for symbols position predictions using the adversarial loss (GAN). In addition, annotations of symbol horizontal positions on the lines of the IAM dataset of handwritten text have been created.
|
|
|
Mobile App - Journal as a Chat
Dorda, Tomáš ; Dobeš, Petr (referee) ; Herout, Adam (advisor)
This application aimed to create a minimalist mobile application for the secure keeping of an electronic diary with the user interface of the instant messaging application. The application is implemented in the Flutter framework and is available for devices with the Android operating system. The work focuses on the design of the application, its implementation, and continuous testing. The result is a journaling application that allows any journal to be locked with its own password, hidden, or even hidden and locked. The application offers a large number of color themes to customize each diary and an original user interface.
|
|
|
Android App for Security Monitoring of Communication
Klepáčková, Karolína ; Dobeš, Petr (referee) ; Herout, Adam (advisor)
This diploma thesis is focused on implementation of application for security monitoring of network communication of other applications in mobile device with Android platform. Provides users information about security risks that may harm his/her privacy or device. It uses a local VPN to tunnel all data sent to the wireless network. These can be linked to an application that has sent them because the Android kernel is derived from the Linux kernel and can be used to retrieve information about established network connections and the application identifier associated with the connection. This mapping allows to get more information about an app that is potentially dangerous for your mobile device.
|
|
|
Reconstruction of AXMAT Testing Station
Gergel, Matej ; Dobeš, Petr (referee) ; Dvořáček, Jiří (advisor)
The master´s thesis deals with reconstruction of RCF test rig AXMAT. Main goal is design new hydraulic load system and his realization. The first part of thesis describes similar experimental machines and their pros and cons. This analysis is base for complete reconstruciton of AXMAT. New hydraulic load system allows dynamic programmable load cycles. Control system with close loop and feedback was created in software Matlab – Simulink. Main frame was reconstructed too. Control and action parts were asseble to station. Output of thesis is function sample.
|
|
|
OCR Trained with Unanotated Data
Buchal, Petr ; Dobeš, Petr (referee) ; Hradiš, Michal (advisor)
The creation of a high-quality optical character recognition system (OCR) requires a large amount of labeled data. Obtaining, or in other words creating, such a quantity of labeled data is a costly process. This thesis focuses on several methods which efficiently use unlabeled data for the training of an OCR neural network. The proposed methods fall into the category of self-training algorithms. The general approach of all proposed methods can be summarized as follows. Firstly, the seed model is trained on a limited amount of labeled data. Then, the seed model in combination with the language model is used for producing pseudo-labels for unlabeled data. Machine-labeled data are then combined with the training data used for the creation of the seed model and they are used again for the creation of the target model. The successfulness of individual methods is measured on the handwritten ICFHR 2014 Bentham dataset. Experiments were conducted on two datasets which represented different degrees of labeled data availability. The best model trained on the smaller dataset achieved 3.70 CER [%], which is a relative improvement of 42 % in comparison with the seed model, and the best model trained on the bigger dataset achieved 1.90 CER [%], which is a relative improvement of 26 % in comparison with the seed model. This thesis shows that the proposed methods can be efficiently used to improve the OCR error rate by means of unlabeled data.
|
guest ::
RSS feed.