|
|
Spracovanie geodát prostredníctvom rozšírenej reality
Kyselica, Ján
This master thesis deals with the design and implementation of an application used for spatial data processing. The first part provides an overview of different augmented reality projects focused on manipulation with spatial data. It also contains review of computer vision libraries. In the methodology, description of all project components is provided. The following part describes in detail the design and implementation of mentioned application for spatial data processing. The proposed implementation is verified on scenario of spatial turn-based game
|
|
|
Rozpoznávaní šachových figurek v partii
Sekanina, Lukáš
SEKANINA, Lukáš. Recognition of chess pieces in game. Bachelor thesis. Brno, 2013. Thesis supervisor Ing. Jan Kolomazník. This thesis describes the design, implementation and testing program for the recognition chess game. The thesis includes the unsuccessful attempts at implementation. Implementation of the program fails to recognize figures using Matlab.
|
|
|
Optimalizace algoritmů pro opakované vyhledání objektu
Juránek, Luboš
The goal of this thesis is to develop a method for repeatedly object detection. This method uses a sequence of images, allowing faster object detection than similar methods working with only one image. Selected problem is solved by creating a demonstration method for object detection by a robot K4 which was created by robotics team PEF MENDELU AiStorm. This demonstration method is later improved to be able to work with sequence of images. This improvement accelerates method by 43 %. The main benefits of the thesis are the possibilities of image processing to object detection, example of the use of a sequence of images to get the same results faster and also a possible solution of computer vision for demonstration system, the robot K4 intended for Ketchup House contest.
|
|
|
Detekce polohy objektů pomocí zpracování obrazu
Koláček, Ladislav
This thesis focuses on development of two methods, which are providing computer vision for robot. The first method is based on Hough transform and the second one is based on machine learning. The chapters in this thesis describe basic methods of image processing and object detection in image, which are closely associated with the development of the earlier mentioned methods. The thesis also includes comparison of processing time and detection rate of the implemented methods. Tests have proven that better results were obtained with the method based on Hough transform.
|
|
|
Řešení vybraných úloh počítačového vidění
Valach, Jiří
The diploma thesis focuses on solving selected machine vision tasks using the LabVIEW development environment and its Vision library. The theoretical part describes selected methods and algorithms in the field of machine vision, as well as used equipment necessary for the preparation of scenes. These are mainly different types of industrial cameras, lenses, filters, and illuminators. Furthermore, the author describes the selected functionality of the Vision library with emphasis on the functions used in the practical part. The focus of the work is the solution of four selected machine vision tasks from the field of industrial practice. Specifically, it involves the detection and control of product quality, determining the occurrence of objects, processing the image taken by a line camera, and quality control of transparent materials. The solution includes not only the design and implementation of software but also the selection of suitable hardware equipment and scene preparation.
|
|
|
Mobile application for detection of skin changes
Awuni, Lawrence Abdullah
Awuni, L.A, Mobile application for detection of skin changes. Diploma thesis. Brno: Mendel University, 2021. This diploma thesis deals with the design and implementation of an android application to detect changes on the skin, leveraging computer vision to argue the difference between images. Its implementation focuses on distinguishes changes in the appearance of birthmarks, moles, and other objects on the skin. This thesis has outlined the requirement for the created application, from design, implantation to deployment
|
|
|
Koncept chytré budovy: monitorování aktivity studentů
Jeřábek, Jiří
The diploma thesis deals with a system for detecting people, faces and activities of Mendel University students. The gathered data are used as a feedback tool for the teacher and may be used to optimize the energy costs of buildings or as a tool for efficient usage of space.
|
|
| |
|
|
Optimal Crop For Photographing Individuals During Sports Activities
Lebedenko, Anastasiia ; Tesařová, Alena (referee) ; Herout, Adam (advisor)
Cílem této bakalářské práce je vyvinout řešení pro zpracování záběrů lidských subjektů za účelem generování videí optimálních rozměrů zaměřených na jednotlivce a eliminujících nadbytečné pozadí. Pomocí modelů počítačového vidění program identifikuje a sleduje lidské pozice ve vstupních videích a poté aplikuje specializovaný algoritmus oříznutí pro generování výstupních snímků. Řešení nabízí možnosti přizpůsobení poměru stran a režimu oříznutí ve výstupním videu. Eliminuje tak nutnost pořizovat více videí, aby byly splněny různé technické nebo estetické požadavky, což umožňuje vytváření různých výstupů z jednoho videa s vysokým rozlišením pomocí předdefinovaných parametrů oříznutí.
|
|
|
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.
|
guest ::
