|
|
A Reduced Neural Network for Classifying the Presence of People in an Image
Stanček, Rastislav ; Rydlo, Štěpán (referee) ; Goldmann, Tomáš (advisor)
Táto práca sa zameriava na tému počítačového videnia, presnejšie, na binárnu klasifikáciu prítomnosti ľudí v obrazových dátach. Cieľom tejto práce je vytvoriť redukovanú neurónovú sieť s využitím metódy knowledge distillation. Klasifikácia a detekcia objektov je výpočtovo náročná operácia. Študentský model vytvorený pomocou knowledge distillation vykazuje ekvivalentnú presnosť, pričom je menší a má vyššiu inferenčnú rýchlosť v porovnaní s učiteľským modelom. Takýto model môže byť interdisciplinárne všestranný a to predovšetkým na koncových zariadeniach, ktoré majú relatívne slabé výpočtové schopnosti.
|
|
|
Prediction of the effect of mutations on protein stability
Rosinská, Monika ; Martínek, Tomáš (referee) ; Musil, Miloš (advisor)
Cílem této práce bylo vytvořit klasifikační model, který dokáže rozlišovat mutace proteinu jako stabilizující či destabilizující. Model byl implementován pomocí konvolučních neuronových sítí s využitím předtrénované sítě ResNet50V2 a datové sady založené na FireProtDB. Model dokáže predikovat vliv mutací na stabilitu proteinu s přesností 0.7 a zároveň poskytuje informaci o míře jistoty dané predikce.
|
|
|
A modern approach to measuring antibiotic susceptibility of microbial cultures using machine learning
Lepík, Jakub ; Burget, Radim (referee) ; Čičatka, Michal (advisor)
The bachelor's thesis focuses on antibiotic susceptibility testing (AST), specifically enhancing and automating the assessment of the disk diffusion method using machine learning and object detection architectures. Thanks to the TensorFlow development platform and extensive dataset, on which custom detection models like EfficientDet were trained, processing a wide range of input data is enabled. This brings the possibility of using mobile devices alongside traditional laboratory equipment when evaluating this method. By employing additional image processing techniques and the OpenCV library, a custom algorithm for measuring the size of inhibitory zones was developed, which, along with the detection models, is integrated within the application module developed by Bruker Daltonics GmbH & Co. KG. This module, created using the ASP.NET platform, is a precise and valuable tool for assisting personnel in microbiological laboratories.
|
|
|
An Application for Analyzing the Resilience of Facial Recognition Algorithms against a Deepfake Image
Kučík, Adam ; Pleško, Filip (referee) ; Goldmann, Tomáš (advisor)
This thesis focuses on creating an application capable of determing whether a given image is deepfake or not. The application is created by using the convolutional neural network vgg-net. Part of the work is to create a Siamese neural network and test if it is suitable for detecting deepfake images. Several configurations of vgg16 and vgg19 networks are created within the thesis. Each configuration contain tables with the success rates of individual models agains our created deepfake dataset. The thesis also includes a section where we discuss deepfake algorithms that are open-source and describe the work with them. The entire application is implemented in Python using the TensorFlow library.
|
|
|
Detekce karet při turnajích v pokru
Kovalets, Vladyslav ; Šilling, Petr (referee) ; Vaško, Marek (advisor)
This bachelor's thesis focuses on the development of an advanced system for automatic recognition and registration of playing cards from video recordings of poker games. The technology of convolutional neural networks, specifically the YOLO network, was chosen as the basic tool. It enables effective identification of cards on the table and in the hands of players even under challenging conditions. The work involved creating an extensive dataset for training and testing the card detector, which achieved a recognition accuracy of 98.7%. An algorithm was designed to minimize detector errors and improve the overall accuracy of the system. The results of the study suggest that the developed system has potential for use in practice.
|
|
|
Potential of neural networks using capsules for medical image processing
Šipula, Samuel ; Vičar, Tomáš (referee) ; Chmelík, Jiří (advisor)
The following master thesis introduces the reader to a relatively new deep learning approach, the capsule neural network. The thesis describes the working principle of capsule networks and compares them with established convolutional networks. Further, the reader is introduced to the use of this technique in medical image processing. The practical part of the paper describes the procedure of learning a capsule network and a reference convolutional network on two datasets. The aim of the thesis is to compare the effect of dataset size on the resulting efficiency of the two types of networks.
|
|
|
Image Inpainting using Deep Learning
Zobaník, Radek ; Kubík, Tibor (referee) ; Šilling, Petr (advisor)
In this thesis, an application was developed for testing and comparing methods for completing missing parts of an image using deep learning, and two methods were trained, pconv with convolutional architecture, and AOT-GAN with GAN architecture. The thesis describes the design of the finished application, its functionality, and important implementation details. A dataset was selected on which the chosen models were optimally trained. Experiments were made on the AOT-GAN model to investigate the impact of the number of AOT blocks in generator on the resulting completed image. All experiments were qualitatively and quantitatively compared. The results showed respectable outcomes when working with natural scenery.
|
|
|
Signature verification using neural network-based algorithms
Čírtek, Petr ; Kiac, Martin (referee) ; Myška, Vojtěch (advisor)
Signature is one of the most used biometrics in banking and contracting therefore is important to verificate signature authenticity. Verification can be done with the help of a forensic specialist or, thanks to the rise of advanced technology, with the help of a computing technology. The purpose of this thesis is to develop methods for signature verification using neural networks for Czech type of signature and to find out if adding manual extracted features to convolutional analysis could improve these methods. Neural networks seek to replicate the functioning of human brain, consisting of input neurons, several hidden layers and output neurons. Neural networks are one of the most popular artificial intelligence technologies for image analysis and classification. The proposed methods in this thesis work on the principles of convolutional networks. The first proposed method consist of three convolutional layers which extract important features from image of signature and pass them to fully connected classifier layer. This determines whether the signature is genuine or forgery. Also for this method there were created two functions which can interpret it's decision-making. The second method, siamese neural network, unlike the first, does not work with signatures independently, but uses a reference signature image to determine authenticity. The basis of this method is to extract features with convolutional analysis from both the reference signature and the signature to be authenticated. These features are then concatenated and passed to the clasificator. A Czech dataset was created to train models that would verify the Czech type of signatures. From the experiments, it was found that the addition of manualy extracted features has the potential to improve the prediction accuracy of methods based on convolutional image analysis. 3 models were trained, which can verify the Czech type of signatures with an accuracy higher than 80 \%, namely: the model of the convolutional neural network method with discrete wavelet transformation feature, which was trained on the Czech dataset, the model of the same method trained on the CEDAR dataset with number of strokes as added feature and a siamese convolutional neural network method model trained on the Czech dataset of signatures with the tri-surface feature.
|
|
|
Segmentation of hyperspectral images of lizards
Kotrys, Kryštof ; Parák, Roman (referee) ; Škrabánek, Pavel (advisor)
Tato diplomová práce se zaměřuje na tvorbu systému pro segmentaci hyperspektrálních fotografií ještěrek žijících na území České republiky. První část práce obsahuje shrnutí existujících metod segmentace obrazu, informací o hyperspektrálním obrazu a konvolučních neuronových sítí. Druhá část práce navrhuje postup pro zpracování dat, které vede k tvorbě segmentovaných masek pro zadanou datovou množinu a také prezentuje získané výsledky.
|
|
|
Object Detection on the i.MX RT Microcontroller
Kravchuk, Marina ; Rozman, Jaroslav (referee) ; Janoušek, Vladimír (advisor)
This work focuses on the use of machine learning, particularly convolutional neural networks, in industrial applications. The course of work involves investigating the implementation of these networks directly on embedded devices, specifically NXP i.MX RT microcontrollers. During the course of the study, materials related to the training and use of neural networks and their optimization for deployment on low power devices were reviewed. Several neural network models were trained and tested, the best of which was used in the final version of the application. The application itself is divided into two parts: one part is written in C/C++ in the MCUXpresso IDE, where the main functionality of the program is implemented, while the other part of the work, i.e. the creation of a graphical user interface to control the program, is done in Python. The result is a functional application for the MIMXRT1170-EVK microcontroller that is able to detect and recognize small colored objects of certain shapes from a predefined data set.
|
guest ::
