|
|
Peer-to-peer Network for Decentralized File Storage and Distributed Task Processing
Mirbauer, Martin ; Zavoral, Filip (advisor) ; Jiráček, Zbyněk (referee)
Peer-to-peer networks allow development of high-performance, scalable and fault-tolerant applications. The most widely used peer-to-peer networks are mostly single-purpose (e.g. file-sharing) and usually lack countermeasures aga- inst sybil nodes; typical task distribution systems are centralized, which limits their performance. In this work we have analyzed the usability of peer-to-peer architecture for task distribution and designed the required infrastructure (file storage and search) including evaluation of possible security features. In order to evaluate the proposed design's usability we have implemented a client ap- plication (node) participating in the network's function, allowing access to files shared by other nodes and task distribution within the network. The resulting application is multi-platform and can be extended with more functionality.
|
|
| |
|
|
Weather Webcam Calibration
Krsička, Ondřej ; Mirbauer, Martin (advisor) ; Šikudová, Elena (referee)
The work deals with the calibration of meteorological web cameras, which can be then used for training of neural networks for the creation of realistic lighting and the genera- tion of realistic clouds in the sky into 3D scenes. A method of calibration using the sun position, a method using the sky appearance and a simple analytical sky model and a method based on neural networks are compared. The first two methods are implemented in Python. A more sophisticated sky model is incorporated into the sky appearance cali- bration method and compared to the original sky model and the sun position calibration. Sun position calibration is the most successful, calibration with the original sky model a little less. The more sophisticated sky model doesn't improve the calibration the way it is used and further improvements are suggested. 1
|
|
|
Generative neural networks for sky image outpainting
Mrázek, Matěj ; Šikudová, Elena (advisor) ; Mirbauer, Martin (referee)
Image outpainting is a task in the area of generative artificial intelligence, where the goal is to expand an image in a feasible way. The goal of this work is to create a machine learning algorithm capable of sky image outpainting by implementing sev- eral recently proposed techniques in the field. We train three models, a tokenizer for converting images to tokens and back, a masked generative transformer for performing outpainting on tokens and a super sampler for upscaling the result, all on a dataset of sky images. Then, we propose a procedure that combines the trained models to solve the outpainting task. We describe the results of training each model and those of the fi- nal algorithm. Our contribution consists mainly in providing a working, open-source implementation including the trained models capable of sky image outpainting. 1
|
|
|
Autoregressive action-conditioned 3D human motion synthesis using latent discrete codes
Waltl, Jan ; Šivic, Josef (advisor) ; Mirbauer, Martin (referee)
V této práci jsme představili novou metodu pro syntézu 3D animace pohybu člověka podmíněné na pevné množině akcí definující pohyb, například "běhání" nebo "předklon". Inspirování úspěchy metod pro generování obrázků z textu na základě diskrétních latentních reprezentacích, jsme úspěšně vyzkoušeli použití těchto metod v kontextu generování pohybu, což je v kontrastu s dosavadními příst upy využívající spojité latentní proměnné. Ve srovnání s dosavadní nejlepší metodou ACTOR, naše metoda není limitována délkou generovaných sekvencí a dokáže plynule navázat na vstupní startovní sekvenci. Autoregressivní generování je omezeno délkou kon textu, což zajišťuje rozumnou rychlost generování. Dále, díky učení ve dvou fázích, budoucí modely se mohou snadno pře dučit na větších datasetech bez označení kategorií a dotrénovat se na konkrétním úkolu. Naši metodu jsme vyhodnotili na UESTC dataset, v metriká ch překování dosavadní metodu ACTOR a generuje animace srovnatelné s datasetem.
|
|
|
HDR image semantic segmentation
Fadeev, Rustam ; Šikudová, Elena (advisor) ; Mirbauer, Martin (referee)
This work attempts to create a pipeline that accepts the high dynamic range (HDR) input in the .exr format, processes it, and feeds it to the deep neural network, which can perform a semantic segmentation task, detecting the sky. Currently, to our knowledge, available semantic segmentation models cannot accept HDR input in the .exr format. Several models that are trained on HDR input are presented and analyzed here. 1
|
|
|
Comparison of deep learning and classical methods for traffic signs detection
Geiger, Petr ; Šikudová, Elena (advisor) ; Mirbauer, Martin (referee)
The goal of this thesis is to explore and evaluate classic and deep neural network computer vision methods in the task of detection position of a level crossing barrier. This thesis is based on an initial detection algorithm using a Stable Wave Detector. The initial algorithm is optimized both in performance and quality of the results. Both is crucial, because the best method should be suitable as a component of the real-time level crossing safety system. Then an another approach is implemented using deep neural networks and optimized in the same manner. Throughout the work several datasets are created for both training and testing of the algorithms. Both approaches are finally evaluated on the same test datasets and the results are compared.
|
|
|
Registration and determination of deformed point cloud errors
Novák, Ondřej ; Blažek, Jan (advisor) ; Mirbauer, Martin (referee)
In this work, we focus on the use of laser slit scanning rigs for 3D reconstruction of the inner parts of a nuclear reactor. Special emphasis is placed on the influence of radial distortion on the quality of 3D reconstruction. We review existing research and available tools used for camera calibration, discuss types of distortion correction models, both polynomial and non-polynomial, present algorithms for parameter estimation, and metrics for evaluation. The aim is to improve 3D point cloud registration of slit scans from multiple views with the goal of high-quality reconstruction. We examine the settings of the GoPro HERO9 Black camera, and our ability to model and correct distortion present in the lenses. We also show how applying delayed distortion correction on extracted image points can result in faster and more dense reconstruction models than early distortion correction of input images. 1
|
|
|
Editor lezeckých tras
Sláma, Tomáš ; Mareš, Martin (advisor) ; Mirbauer, Martin (referee)
With the rising popularity of sport climbing and bouldering, there are increasing incentives to take it to the online space. This thesis does so in a way suitable for regular climbing gyms. A workflow for creating 3D hold and wall models and the accompanying editor has been developed, so that routes can be copied from the real world to the virtual one and be interacted with using software in a more immersive way. 1
|
|
|
An Application for Taking 360° Photographs using Mi Sphere Camera
Cífka, Martin ; Mirbauer, Martin (advisor) ; Horký, Vojtěch (referee)
Xiaomi Mi Sphere is camera that captures 360ř panoramatic photographs. With si- multaneous use of high dynamic range (HDR) capturing, such photographs are used in realistic rendering as a background as well as for creating realistic lighting and light re- flections. The advantage of this camera is that it's lightweight and easily carried, but the official application is available only for Android and iOS devices and it does not support simultaneous use of time-lapse and bracketing for capturing HDR images. Other unoffi- cial applications do not support this functionality either. In this thesis we analyze the communication protocol between camera and the official application and subsequently we develop our own cross-platform application with web-based graphical user interface. In addition to standard capturing, our application supports simultaneous use of bracketing and time-lapse. The created application does not run on the user's device but directly on the camera. This allows us to continue shooting even after client disconnects from the application. 1
|
guest ::
