Národní úložiště šedé literatury Nalezeno 32 záznamů.  1 - 10dalšíkonec  přejít na záznam: Hledání trvalo 0.01 vteřin. 
Vehicle Location Detection and Distribution from Camera Images
Stryk, Filip ; Götthans, Jakub (oponent) ; Götthans, Tomáš (vedoucí práce)
This bachelor's thesis deals with vehicle location detection and tracking. Basic principles of deep learning and convolutional networks are presented. Deep learning based object detectors are described with a focus on YOLO and then compared in terms of accuracy and speed. A system for vehicle location detection and tracking using YOLOv4-tiny and SORT is designed, implemented and evaluated.
Design of adaptive wireless video and data transmission
Lorenc, Tomáš ; Nykodým, Jiří (oponent) ; Götthans, Tomáš (vedoucí práce)
The aim of this thesis is to build a device that will be able to establish a wireless transmission of a real-time video stream and send the video to a computer where it will be displayed. The device is powered by the NVIDIA Jetson Nano and video is streamed wirelessly through a WiFi interface. In theoretical part described the problem of video transmission and video compression. The used codecs are h264, h265, and VP8. A subjective test to find the minimum acceptable value of video quality was performed. Measurements have been made to find the best codec for real-time video streaming. Measurements were also performed in real conditions. In conclusion, the results of the measurements and the selection of the best codec for real-time video streaming are commented.
Optimizing neural network architecture for EEG processing using evolutionary algorithms
Pijáčková, Kristýna ; Maršálek, Roman (oponent) ; Götthans, Tomáš (vedoucí práce)
This thesis deals with an optimization of neural network hyperparameters for EEG signal processing using evolutionary algorithms. The incorporation of evolutionary optimization can reduce reliance on human intuition and empirical knowledge when designing neural network and can thus make the process design more effective. In this work, a genetic algorithm was proposed that is suitable for hyperparameters optimization as well as neural architecture search. These methods were compared to a benchmark model designed by an engineer with expertise in iEEG processing. Data used in this work are classified into four categories and come from St. Anne's University Hospital (SAUH) and Mayo Clinic (MAYO) and were recorded on drug-resistant epileptic patients undergoing pre-surgical examination. The results of the neural architecture search method were comparable with the benchmark model. The hyperparameter optimization improved the F1 score over the original, empirically designed, model from 0.9076 to 0.9673 for the SAUH data and 0.9222 to 0.9400 for the Mayo Clinic data. The increased scores were mainly due to the increased accuracy of the classification of pathological events and noise, which may have further positive implications in applications of this model in seizure and noise detectors.
Model Ensembeling: A simple way of improving model performance for chromosome classification
Pijáčková, Kristýna ; Gotthans, Tomáš ; Gotthans, Jakub
This paper deals with chromosome classification via convolutional neural networks and model ensembling. Chromosome classification is a part of a procedure in karyotyping, where the chromosomes should be paired and ordered so that they are prepared for inspection of abnormalities. Model ensembling was used as a technique to improve overall classification accuracy by using all of the trained models. We achieved 94.8 \% accuracy for a Q-band BioImlab dataset and 97.48 \% for a G-band chromosome CIR dataset.
Digital Predistorters with Low-Complexity Adaptation
Král, Jan ; Springer, Andreas (oponent) ; Roblin, Patrick (oponent) ; Götthans, Tomáš (vedoucí práce)
Modern communication systems often require digital predistorters (DPDs), advanced signal-processing units, to satisfy stringent demands on transmitter linearity and efficiency. Nevertheless, DPD significantly increases the hardware and computational complexity of transmitters, which leads to increased power consumption and expenses. Therefore, we propose methods to achieve lower hardware and computational complexity of DPD adaptation. The principle of real-valued feedback samples allows for saving one of two originally-needed feedback analogue-to-digital converters (ADCs), which implies reduced transmitter complexity and power consumption. Furthermore, the hardware and computational complexity can be reduced if the feedback samples for the DPD adaptation are undersampled and carefully selected. The proposed techniques select samples based on histograms and can reduce the required number of feedback samples to a few tens. The provided analyses show approximately 400-times reduced computational complexity achieved by the sample selection and 40-times reduced power consumption of the undersampling feedback ADCs. The real-valued feedback, its undersampling, and sample selection constitute fundamental principles of the proposed DPD adaptation with a level-crossing ADC, which is realised by a simple comparator. Replacing the conventional ADCs with a comparator significantly reduces the design complexity and power consumption. All the proposed and described techniques are accompanied by simulations, usually confirmed by measurements on real hardware, and compared with state-of-the-art methods. The final discussion analyses the limitations, usability and advantages of the proposed techniques. It shows that reducing complexity might not be universally applicable and all the design constraints and specifications must be carefully assessed.
Komunikační síť pro rojové létání bezpilotních letounů
Rajm, Jan ; Götthans, Tomáš (oponent) ; Janoušek, Jiří (vedoucí práce)
Tato bakalářská práce se zabývá vytvořením robustní komunikační sítě pro rojové létání bezpilotních letadel a jejím testováním. V první části práce jsou představeny požadavky na komunikační síť a jsou popsány vybrané komunikační standardy, které jsou schopné přenášet telemetrická data. Ve druhé části je popsán software potřebný pro přenos telemetrických dat a pro řízení bezpilotních letadel. Ve třetí části je představena navržená komunikační síť pro roj bezpilotních letadel. Jsou prezentovány její možnosti a odlišnost od sítí založených na jiných technologiích. V této části je také popsán postup jejího sestavení. Čtvrtá část práce se zabývá testováním a vyhodnocováním výsledků.
Bluetooth lokalizace s nízkou spotřebou energie
Šlígl, Jan ; Nykodým, Jiří (oponent) ; Götthans, Tomáš (vedoucí práce)
Tato diplomová práce se zabývá návrhem lokalizačního zařízení v Bluetooth 5.2 s nízkou spotřebou. Lokalizace je provedena pomocí vytvořených zařízení tvořících beacony a navržené aplikace pro mobilní telefon. Dále také návrhem více druhů antén s různými směrovými charakteristikami, ziskem a velikostmi. Návrh bere v potaz především nízkou spotřebu energie, kompaktnost a nezávislost zařízení.
Radiový DAB přijímač
Pišťák, Ondřej ; Götthans, Tomáš (oponent) ; Dřínovský, Jiří (vedoucí práce)
Práce je zaměřena na návrh a konstrukci softwarově definovaného rádia, které umí přehrávat standard DAB+. Zpracování signálu zachyceného pomocí antény zajišťuje USB přijímač RTL-SDR BLOG s 8 – bitovým A/D převodníkem RTL2832U a platforma Raspberry Pi 3 (Model B). Práce je také zaměřena na návrh a realizaci ovládacích prvků k aplikaci pro jednoduché přepínání stanic pomocí tlačítek na ovládacím panelu. Pro zobrazení právě přehrávané stanice slouží připojený LCD display.
Radio Modulation Recognition Networks
Pijáčková, Kristýna ; Maršálek, Roman (oponent) ; Götthans, Tomáš (vedoucí práce)
The bachelor thesis is focused on radio modulation classification with a deep learning approach. There are four deep learning architectures presented in the thesis. Three of them use convolutional and recurrent neural networks, and the fourth uses a transformer architecture. The final number of parameters of each model was considered during the design phase, as it can have a big impact on a memory footprint of a deployed model. The architectures were written in Keras, which is a software library, which provides a Python interface for neural networks. The results of the architectures were additionally compared to results from other research papers on this topic.
Adaptation of digital predistorter to linearize amplifiers using comparator
Jagla, Lukáš ; Ayöz, Suat (oponent) ; Götthans, Tomáš (vedoucí práce)
This master’s thesis presents a development of a new hardware implementing a comparator in the feedback path of DPD systems. A new architecture is proposed and selected features are verified by simulations. Subsequently, the suitable components are selected for high-speed performance and an acquisition module is proposed. A 4-layer PCB is well designed, manufactured, and prepared for further work. Afterwards, an appropriate firmware is developed for signal transmission and data acquisition. The obtained results serves for the evaluation of the proposed architecture and for its future implementation in real DPD systems.

Národní úložiště šedé literatury : Nalezeno 32 záznamů.   1 - 10dalšíkonec  přejít na záznam:
Viz též: podobná jména autorů
3 Gotthans, Tomáš
Chcete být upozorněni, pokud se objeví nové záznamy odpovídající tomuto dotazu?
Přihlásit se k odběru RSS.