|
|
Artificial intelligence on nVIDIA Jetson platform
Batelka, Lukáš ; Kozovský, Matúš (referee) ; Blaha, Petr (advisor)
The aim of this bachelor thesis is to design, train and implement an artificial neural network in an NVIDIA Jetson Nano embedded device. The first part of the thesis describes the current state of the art of implementing artificial intelligence in embedded devices. The following section describes the tools for developing artificial neural networks and the possibilities of implementing them in a Jetson Nano device. These tools are further used in the thesis to create and train an artificial neural network to detect a fault in preprocessed measurement data on a synchronous electric motor. Finally, the optimization of the trained neural network is described. The achieved results are summarized in the conclusion of the paper.
|
|
|
Detection and Classification of Damage in Fingerprint Images Using Neural Nets
Vican, Peter ; Drahanský, Martin (referee) ; Kanich, Ondřej (advisor)
The aim of this diploma thesis is to study and design experimental improvement of the convolutional neural network for disease detection. Another goal is to extend the classifier with a new type of detection. he new type of detection is damage fingerprint by pressure. The experimentally improved convolutional network is implemented by PyTorch. The network detects which part of the fingerprint is damaged and draws this part into the fingerprint. Synthetic fingerprints are used when training the net. Real fingerprints are added to the synthetic fingerprints.
|
|
|
Object clasification based on its topology change using image processing
Zbavitel, Tomáš ; Věchet, Stanislav (referee) ; Krejsa, Jiří (advisor)
The aim of the present work is to select a suitable object classification method for the recognition of one-handed finger alphabet characters. For this purpose, a sufficiently robust dataset has been created and is included in this work. The creation of the dataset is necessary for training the convolutional neural network. Further more, a suitable topology for data classification was found. The whole work is implemented using Python and the open-source library Keras was used.
|
|
| |
|
|
Detection of Boxes in Image
Soroka, Matej ; Zlámal, Adam (referee) ; Herout, Adam (advisor)
The aim of this work is to experiment and evaluate algorithms with different approaches to computer vision in order to automatically detect boxes-blocks in the image. To this end, neural network-based approaches were used in the solution. Experiments were performed with classification using our own data set, classification using our own convolutional neural network, detection using a window, YOLO detector and in the final iteration the use of U-net network for detection of boxes in the image.
|
|
|
Segmentation of cardiac muscle images acquired using confocal microscopy
Kadlec, Filip ; Shehadeh, Mhd Ali (referee) ; Škrabánek, Pavel (advisor)
Automatizace získávání a zpracování dat je dnes běžnou záležitostí jak v mikroskopii tak v počítačovém vidění. Ke klasifikaci a lokalizaci objektů zájmu (v tomto případě kardiomyocytů) lze užít segmentaci. V tomto konkrétním případě byla aplikována sémantická segmentace za užití hlubokých neuronových sítí jakožto hlavního prostředku k provedení zmíněné úlohy a byl vytvořen software umožňující jak zpracování neoznačených dat, tak trénování modelů neuronových sítí na označených datech. Tato práce krátce hovoří o optické mikroskopi, detailně popisuje segmentaci a hluboké učení a na závěr poskytuje popis procesu od přípravy dat, přes implementaci a trénování neuronových sítí, k vytvoření konečného softwaru. Tento software usnadní a zefektivní práci výzkumníků poskytnutím pouze relevantních dat pro výzkum, pomůže automatizovat sběr mikroskopických snímků a s menším upravením může být aplikován na další obdobné segmentační úlohy.
|
|
|
Machine Translation Using Artificial Neural Networks
Holcner, Jonáš ; Beneš, Karel (referee) ; Szőke, Igor (advisor)
The goal of this thesis is to describe and build a system for neural machine translation. System is built with recurrent neural networks - encoder-decoder architecture in particular. The result is a nmt library used to conduct experiments with different model parameters. Results of the experiments are compared with system built with the statistical tool Moses.
|
|
|
Reinforcement learning for solving game algorithms
Daňhelová, Jana ; Uher, Václav (referee) ; Kolařík, Martin (advisor)
The bachelor thesis Reinforcement learning for solving game algorithms is divided into two distinct parts. The theoretical part describes and compares the fundamental methods of reinforcement learning with special attention to the methods of active learning – Q-learning and deep learning. In the practical part the deep q-learning technique is chosen for testing and applied to the case of the Snake game. The results are presented in the form of program written in Python programming language, which consists of the game environment created in PyGame, the model of convolutional neural network designed in Keras and agent playing the game. As an output of the program there are several types of datasets in CSV format. The gained data containing the values of parameters like number of epochs, accuracy, loss or the amount of the reward can later be used for further processing.
|
|
|
Recognition of Vehicle Class in Image
Čabala, Roman ; Kodym, Oldřich (referee) ; Špaňhel, Jakub (advisor)
The goal of this bachelor thesis is to recognize the type of vehicle from the image using neural networks. Vehicles are divided into 6 types, namely a car, a small van, a van, a mini truck, a truck and a bus. The data set was picked from videos that record the trajectory of the vehicles. Subsequently, an image annotation tool was built. The following architectures were used for network training: VGG16, ResNet50, Xception, InceptionResNet-v2. The result of the work is a comparison of architectures. All architectures were trained and achieved a result above 90%.
|
|
|
Polygonal Mesh Segmentation
Bezděčík, Ladislav ; Polášek, Tomáš (referee) ; Španěl, Michal (advisor)
This bachelor's thesis deals with the issues of segmentating 3D models of human jaws. It analyzes currently used methods and proposes, implements and tests possible improvement to these methods from user perspective. The proposal consists of using neural networks for topology recognition on jaw models, and possibly combining this topology with currently used segmentation methods. This thesis also analyzes and implements the possibility of automated expnansion of 3D model datasets converted to depth maps, used for neural network training.
|
guest ::
