|
|
Evolutionary Design of Neural Networks
Kastner, Jan ; Hurta, Martin (referee) ; Sekanina, Lukáš (advisor)
The thesis deals with the implementation of a problem-solving method for the automated design of convolutional neural networks (CNN) architectures. The optimization of two fundamental and often conflicting characteristics, the number of parameters and the quality of CNN classification, is performed using a multi-criteria optimization genetic algorithm (NSGA-II). To encode this problem, the Cartesian genetic programming (CGP) technique is used, which enables the wide range of CNN architectures to be represented, and at the same time, the searched area can be appropriately limited by parameterization. Experiments were performed on the MNIST dataset to understand the effect of population size on the quality of the resulting solution. It is also evident from the results of the experiments that the quality of the architectures found can compete with already established models. This is therefore an alternative approach that does not require human intervention compared to manual design.
|
|
|
The use of deep neural networks for the evaluation of metallographic cross-sections
Semančík, Adam ; Mendřický, Radomír (referee) ; Hurník, Jakub (advisor)
Táto diplomová práca skúma aplikáciu hlbokých neurónových sietí pre vylepšenie hodnotenia metalografických výbrusov pre materiály vyrobené pomocou aditívnej výroby. Zameriava sa na dve pokročilé techniky spracovania obrazu: sémantickú segmentáciu a super-rozlíšenie obrazu. Na sémantickú segmentáciu bola použitá architektúra U-Net pre klasifikáciu defektov, ako sú dva typy pórov. Okrem toho bol použitý model SRGAN (Super-Resolution Generative Adversarial Network) pre zvýšenie rozlíšenia obrazu, čo potenciálne zlepšuje presnosť segmentácie. Výskum hodnotí, či model trénovaný na AlSi10Mg môže dostatočne dobre vyhodnocovať materiály Cu99 a Ti6Al4V. Zároveň hodnotí vplyv super-rozlíšenia na výkonnosť segmentácie. Výsledky ukázali, že zatiaľ čo model segmentácie dosahoval dobré výsledky na AlSi10Mg, generalizácia na iné materiály vyžaduje diverzifikovanejšie tréningové dáta. V dôsledku výpočtových obmedzení zostáva kombinovaný efekt super-rozlíšenia a segmentácie nejednoznačný, čo naznačuje potrebu ďalšieho výskumu s výkonnejšími výpočtovými zdrojmi.
|
|
|
Human-machine collaboration - using speech processing
Kisler, Štěpán ; Hůlka, Tomáš (referee) ; Juříček, Martin (advisor)
This bachelor's thesis focuses on the design and implementation of a voice control system for the UR3 CB series collaborative robot from Universal Robots, aiming to simplify human-robot interaction. The introduction provides an overview of collaborative robotics, including its history, successful applications, and the possibilities of programming collaborative robots. Additionally, it explores speech recognition technology, covering its applications, history, and methods. The practical section compares existing speech recognition systems and selects the most suitable one for robot voice control. It also details the development of a voice control program in Python and the testing of the entire system, both in simulation and real-world conditions in a robotics laboratory.
|
|
|
Stereo Reconstruction with Deep Neural Networks
Letanec, Richard ; Herout, Adam (referee) ; Španěl, Michal (advisor)
The aim of this thesis is to design and train a neural network model capable of estimating a disparity map from a pair of images. It will then be possible to create a depth map and point cloud from the estimated disparity map. Such a process is called stereo reconstruction. Solving this task consists of two steps -- choosing a suitable dataset and choosing a suitable neural network architecture. In my work, I compared two neural network architectures that I trained on the DrivingStereo dataset, consisting of paired images photographed from the roof of a car, and retrained and evaluated on the KITTI 2015 dataset, consisting of images of the same type. As the first neural network architecture, I chose ES-Net, which uses an approach based on a sequence of residual blocks and convolutional layers. As the second architecture, I chose CREStereo, which uses an iterative approach based on recurrent layers to predict the disparity map. In all benchmark tests, the CREStereo architecture achieves better accuracy.
|
|
|
Face Anti-Spoofing with Out-of-distribution Detection
Češka, Petr ; Vaško, Marek (referee) ; Špaňhel, Jakub (advisor)
This thesis aims to improve the accuracy of Vision Transformer-based face anti-spoofing models in detecting presentation attacks. The thesis uses out-of-distribution detection to filter out images that are too different from the training data, referred to as in-distribution. It examines how successful different methods are in identifying different data distributions, and how the filtering of out-of-distribution data based on these methods affects the accuracy of the model. Using the relative Mahalanobis distance, an AUROC of 97.6% can be achieved when distinguishing between in-distribution and out-of-distribution data. Filtering out images that should not be classified increases the accuracy of all tested models to over 99.9%. This can provide an additional layer of security for applications against face spoofing attacks.
|
|
|
Video Denoising Using Deep Learning
Naumenko, Maksim ; Hradiš, Michal (referee) ; Španěl, Michal (advisor)
V éře digitálních multimédií kvalita videoobsahu významně ovlivňuje uživatelský zážitek a výkon systému, zejména v oblastech, jako je zábava a zpracování videa a obrazu. Tato práce se zabývá přetrvávajícím problémem šumu ve videu, který zhoršuje jeho kvalitu, a to pomocí pokročilých technik hlubokého učení. Nejprve jsou přezkoumány tradiční přístupy k odstraňování šumu ve videu, aby bylo možné nastínit základní koncepty denoisingu. Následně jsou studovány dva referenční modely, FastDVDNet a ViDeNN, za účelem seznámení se s architekturami neuronových sítí. Hlavním výsledkem této práce je vývoj robustního systému pro odstraňování šumu ve videu, který je založen na architektuře UNet inspirované těmito referenčními modely. V průběhu práce jsou vysvětleny, implementovány a vyhodnoceny navrhované modely UNet Baseline, ResUNet a ResUNet Temporal, aby byla prokázána jejich účinnost v odstraňování šumu ve videu.
|
|
|
Deep Neural Networks Approximation
Stodůlka, Martin ; Mrázek, Vojtěch (referee) ; Vaverka, Filip (advisor)
The goal of this work is to find out the impact of approximated computing on accuracy of deep neural network, specifically neural networks for image classification. A version of framework Caffe called Ristretto-caffe was chosen for neural network implementation, which was extended for the use of approximated operations. Approximated computing was used for multiplication in forward pass for convolution. Approximated components from Evoapproxlib were chosen for this work.
|
|
|
Reinforcement Learning for RoboCup
Bočán, Hynek ; Škoda, Petr (referee) ; Smrž, Pavel (advisor)
Goal of this thesis is creation of artificial intelligence capable of controlling robotic soccer player simulated in SimSpark environment. Agent created is expanding capabilities of existing third party agent which provides set of basic skills such as localization on the field, dribbling with the ball and omnidirectional walk. Responsibility of the created agent is to pick the best action based current state of the game. This decision making was implemented using reinforcement learning and its method Q-learning. State of the game is transformed into 2D picture with several planes. This picture is then analyzed using deep convolution neural network implemented using C++ and DeepCL library.
|
|
| |
|
|
Improving Bots Playing Starcraft II Game in PySC2 Environment
Krušina, Jan ; Škoda, Petr (referee) ; Smrž, Pavel (advisor)
The aim of this thesis is to create an automated system for playing a real-time strategy game Starcraft II. Learning from replays via supervised learning and reinforcement learning techniques are used for improving bot's behavior. The proposed system should be capable of playing the whole game utilizing PySC2 framework for machine learning. Performance of the bot is evaluated against the built-in scripted AI in the game.
|
guest ::
