National Repository of Grey Literature 21 records found  1 - 10nextend  jump to record: Search took 0.01 seconds. 
Capturing of Detailed and Very Large Photograph and Localization Within
Dubovec, Pavol ; Vaško, Marek (referee) ; Herout, Adam (advisor)
The goal of this work was to create a large image and a new technique to localize the photo in the larger image to increase the speed and accuracy of conventional methods. The proposed technique uses CNN architecture to extract embeddings from the queried image which will be used to search the database of embeddings from the large photo. Two models have been trained on a large dataset: based on classification (CE) and distance (triplet) Conventional methods were used to determine the location of the images and to generate the large image. A database of embeddings was created by partitioning the large image using the trained model. The database is searched for the K-nearest embeddings of the cutouts of the query image. These embeddings are generated by dividing the query photo into the same size parts as the CNN inputs. The optimal homography model is determined by random selection based on the positions of the four query image cutouts and their corresponding positions in the big picture. The homography model with the lowest harmonic mean of the embedding distance is selected as the final position. The homography is optimized using template matching where possible. The method shows sufficient accuracy and high speed on test datasets. The best model achieved a top-1 accuracy of 97.71% and a top-3 accuracy of 99.67%. Future research will investigate the performance of the method under increasing surface heterogeneity, the possibility of automating video retrieval to obtain a large dataset with photos, and its effectiveness in locating photos when conventional methods fail.
Detekce typu a bodového ohodnocení kartiček ve hře Hobiti
Hlinský, Martin ; Kohút, Jan (referee) ; Vaško, Marek (advisor)
This thesis aims to create a card detector that can train a model that can detect the score of a card and its type using the synthetic generation of the dataset. The YOLOv8 model is used for training. The first step is to take pictures of the cards, which then go through a pre-processing stage so they do not contain background and are aligned. These pre-processed card images are combined with photos from other datasets in a generator that randomly translates, rotates, and otherwise simulates photos of possible card placements. This generator’s output is roughly 50 000 annotated images in the case of the Hobiti game, but different dataset sizes and pre-trained weights are compared in the experiments. The latest generation of trained detectors was validated on a real dataset for unbiased testing, and the most accurate model trained on purely synthetic datasets achieved precision up to 81.5 % according to the 50 metric. It is then possible to implement, for example, a point counter on the final detector, a prototype of which is also described in this paper.
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.
Automatická vizuální podpora pro Q-řazení
Kán, Dávid ; Hradiš, Michal (referee) ; Vaško, Marek (advisor)
This bachelor thesis deals with the integration of Q-sorting and computer vision methods for object detection. The goal of the work is to create a program that, with the help of~visual support, will facilitate the process and at the same time prevent errors in Q-sorting. Furthermore, the work deals with the creation of~a suitable data set for training the model and for experiments, which takes into account the way the cards are laid out and the~environment. The implemented program takes the form of a console application and is written using the Python programming language. The program uses YOLOv8 to detect objects and uses Pero OCR to retrieve text from cards. Using the created test set, experiments were performed on the trained model and the program was tested.
Adversarial Attacks on AI Algorithms and Their Prevention
Gregorová, Jana ; Vaško, Marek (referee) ; Herout, Adam (advisor)
Bezpečnost AI a útoky na umělou inteligenci představují komplexní a dosud nedostatečně prozkoumanou problematiku. Cílem této práce je nabídnout ucelený přehled klíčových metod a možných vysvětlení útoků na AI a obran proti nim, aby se toto téma stalo přístupnějším a srozumitelnějším pro širší publikum, a tím usnadnit hlubší zkoumání a porozumění těmto útokům. Tato práce zahrnuje výběr metod pro vysvětlení jednotlivých klasifikačních rozhodnutí klasifikátorů hlubokého učení (Explainable AI: XAI) a jejich aplikaci při analýze rozhodovacího procesu klasifikátorů během útoků. K usnadnění vytváření dalších experimentů, monitorování útoků na AI a hledání možných vysvětlení byl navíc vyvinut skript, který tento proces zjednodušuje. Tento skript je součástí této práce a je poskytnut na přiloženém médiu.
Identifikace člověka podle fotografie dlaně / hřbetu ruky
Štanga, Miroslav ; Vaško, Marek (referee) ; Herout, Adam (advisor)
This work focuses on using contrastive self-supervised learning method for creating model of deep learning intended for person recognition based on hand photographs. The paper outlines fundamentals of machine learning, utilized tools and dataset. The method was developed using PyTorch library. The proposed model draws inspiration from the SimCLR architecture and its use of contrastive representation learning. The proposed approach utilizes the triplet loss function for optimization. Then the optimization process is described and impact of individual hyperparameters on the model´s accuracy is compared. The resulting model was trained on 1696 hand photos and achieves 98% accuracy on validation set. The accuracy achieved using self-supervised methods is higher than the accuracy achieved using supervised methods.
Mobilní aplikace pro podporu trénování silových sportů
Košina, Simon ; Vaško, Marek (referee) ; Juránek, Roman (advisor)
The aim of this work is to create a mobile application for Android devices that provides athletes with real-time feedback during strength training in the form of velocity metrics for individual repetitions within a set of a certain exercise. Velocity based training is becoming increasingly popular both in practical applications and in research, where it has been demonstrated that these objective metrics can be used to estimate the intensity of a given set. The resulting application utilizes machine learning methods to detect weights plates loaded on a barbell in frames coming from the mobile device's camera and tracking their movement trajectory. Known size of the weight plates is used to calibrate the travelled distance. The algorithm operates in real-time, providing users with feedback during exercise sessions in the form of an auditory signal when a predefined threshold of selected velocity metric is reached.
Automated Metadata Extraction From Document Images
Křivánek, Jakub ; Vaško, Marek (referee) ; Kohút, Jan (advisor)
This Bachelor thesis addresses the problem of extracting structured data from scans of documents from Czech libraries. The aim of the thesis is to simplify the time-consuming manual process for librarians. I focused on creating datasets from documents of Czech libraries and on detecting metadata on these datasets. I created one dataset for books and another for periodicals. Detection was performed by classifying lines read from the documents. This utilized a fully connected neural network and a network employing a Transformer Encoder. The second method of metadata detection is based on object detection in document scans using the YOLOv8 model. Detection using the fully connected neural network achieves an F1 score of 0.83 on the book dataset and 0.78 on the periodicals dataset. The Transformer Encoder network achieves F1 scores of 0.84 on the book dataset and 0.59 on the periodicals dataset. The YOLO model achieves an F1 score of 0.86 (confidence at 0.549) on the book dataset and 0.7 (confidence at 0.336) on the periodicals dataset.
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.
Podpora pěstování ve městech pomocí webové aplikace
Čelakovský, Alexandr ; Vaško, Marek (referee) ; Bažout, David (advisor)
The goal of this bachelor thesis is design and implementation of a web application for a smart greenhouse with elements of gamification. The main parts of the system include rental of flowerbeds in individual greenhouses and a smart marketplace that is fully automated. The backend is implemented in the Django framework and the frontend is written in React. Emphasis was placed on the greatest possible automation and simplicity of the user interface. The result of the work is a web application that meets the requirements. Tests have shown that the final solution is functional and user-friendly.

National Repository of Grey Literature : 21 records found   1 - 10nextend  jump to record:
See also: similar author names
6 Vasko, Martin
6 Vaško, Martin
12 Vaško, Michal
Interested in being notified about new results for this query?
Subscribe to the RSS feed.