Defect inspection is one of the most critical tasks in the industry as it can reduce risks of production stops and assure quality control. In recent years, multiple industries have been adopting computer vision systems, especially based on deep learning techniques, as their main detection methods to improve efficiency, reduce risks and human resources, and enhance real-time performance. However, its adoption in the industry is still limited by the labor-intense and time-consuming process of collecting high-quality custom training datasets. At the same time, many industries have access to the CAD models of the components they want to detect or classify as part of the design process. Taking this into account, in the present work, we analyze the performance of various image classification models to visually detect defects in production. Our method systematically generates synthetic datasets from CAD models using Blender to train neural networks under different settings. The proposed method shows that image classification models benefit from a diversity of the range of defect values during training but struggle to identify low-resolution defects, even for state-of-the-art architectures like Vision Transformer and ConvNext or SqueezeNet, which proved to have comparable performance to these networks. Similarly, adding background, texture, and camera pose to training examples provides more contextual information to image classification models but does not necessarily help them detect the defects accurately. Finally, we observed that using a unique tolerance value for all flange pipe sizes can negatively impact the detection accuracy because, for larger pipe flanges, minor defects are not as perceptible as for small flanges.

Šiame darbe tiriamos trimatės Blender aplinkos panaudojamumo garsinių signalų modeliavimui galimybės. Darbe aprašyta programos vartotojo sąsaja, vidinė architektūra, ištirtos vidinio programavimo galimybės ir pateikti tyrimų rezultatai. Naudojant Python programavimo kalbą buvo sukurta garsinių signalų analizės ir vizualizavimo programa SigBlender, veikianti Blender aplinkoje. Programa naudoja beveik periodinių garso signalų periodų išskyrimo algoritmą, kuris buvo sukurtas ir realizuotas darbo metu. Naudojant tekstinį redaktorių Notepad++ sukurta nauja programavimo aplinka, integruojanti Python ir Blender programas. Darbe pateikiami trimačiai garso signalų ir jų amplitudinių charakteristikų grafikai, demonstruojantys trimatės grafikos privalumus garsinių signalų vizualizavime. / This work studies feasibility of a three-dimensional environment, called Blender, for sound signal modeling. It describes user interface and architecture of the environment and analyses capabilities of internal programming. During the work a program for a sound signal analysis and visualization was developed. It was written using Python programming language and was called SigBlender. For extraction of periods of almost periodical sound signals SigBlender uses an algorithm which was developed and implemented during this work. By using the text editor Notepad++ a new programming environment was created and used for coding. Charts of sound signals and their spectral characteristics presented throughout this work illustrate the benefits of visualizing sound signals in a three-dimensional space.

Sistemas de Transportes Inteligentes representam um imenso impacto social e aos poucos tem modificado o paradigma de mobilidade atual. Desde a década de 80, veículos autônomos vêm sendo desenvolvidos e estudados pela comunidade científica e hoje atrai o interesse de grandes empresas automobilísticas. Esses sistemas têm como objetivo a redução do número de acidentes de trânsito, aumento da eficiência dos transportes e inclusão social, sendo que neste contexto surge o projeto CaRINA. Através do desenvolvimento de uma plataforma robótica móvel pretende-se realizar a navegação completamente autônoma em ambiente urbano. Contudo, os experimentos realizados com a plataforma real são custosos, demorados e perigosos. A logística dos testes é complexa, uma vez que necessitam de local apropriado e disponibilidade de recursos. Portanto, o objetivo deste trabalho é desenvolver uma arquitetura de simulação para veículos autônomos que seja capaz de realizar experimentos em laboratório e facilite a portabilidade dos programas desenvolvidos em simulação para o veículo real. A flexibilidade da arquitetura do simulador também permite realizar experimentos utilizando múltiplos veículos. / Intelligent Transportation Systems represent a huge social impact and has gradually modified the current mobility paradigm. Since the1980s, autonomous vehicles have been developed and investigated by the scientific community and today attracts the interest of major automotive companies. These systems aims to reduce the number of traffic accidents, increase the transports efficiency and social inclusion, in this context the CaRINA project started. Through the development of amobile robotic platformis intended to performa fully autonomous navigation in urban environments. However, the experiments with the platform are costly, time-consuming and dangerous. The logistics of the tests are complex, since that require the appropriate location and availability of resources. Therefore, the purpose of this work is to develop a simulation architecture for autonomous vehicles to be able to perform experiments in the laboratory and to facilitate the portability of programs developed in simulation to the real vehicle. The flexibility of the simulator architecture also allows to perform experiments using multiple vehicles.

Sistemas de Transportes Inteligentes representam um imenso impacto social e aos poucos tem modificado o paradigma de mobilidade atual. Desde a década de 80, veículos autônomos vêm sendo desenvolvidos e estudados pela comunidade científica e hoje atrai o interesse de grandes empresas automobilísticas. Esses sistemas têm como objetivo a redução do número de acidentes de trânsito, aumento da eficiência dos transportes e inclusão social, sendo que neste contexto surge o projeto CaRINA. Através do desenvolvimento de uma plataforma robótica móvel pretende-se realizar a navegação completamente autônoma em ambiente urbano. Contudo, os experimentos realizados com a plataforma real são custosos, demorados e perigosos. A logística dos testes é complexa, uma vez que necessitam de local apropriado e disponibilidade de recursos. Portanto, o objetivo deste trabalho é desenvolver uma arquitetura de simulação para veículos autônomos que seja capaz de realizar experimentos em laboratório e facilite a portabilidade dos programas desenvolvidos em simulação para o veículo real. A flexibilidade da arquitetura do simulador também permite realizar experimentos utilizando múltiplos veículos. / Intelligent Transportation Systems represent a huge social impact and has gradually modified the current mobility paradigm. Since the1980s, autonomous vehicles have been developed and investigated by the scientific community and today attracts the interest of major automotive companies. These systems aims to reduce the number of traffic accidents, increase the transports efficiency and social inclusion, in this context the CaRINA project started. Through the development of amobile robotic platformis intended to performa fully autonomous navigation in urban environments. However, the experiments with the platform are costly, time-consuming and dangerous. The logistics of the tests are complex, since that require the appropriate location and availability of resources. Therefore, the purpose of this work is to develop a simulation architecture for autonomous vehicles to be able to perform experiments in the laboratory and to facilitate the portability of programs developed in simulation to the real vehicle. The flexibility of the simulator architecture also allows to perform experiments using multiple vehicles.

This text was written as part of a master's thesis. It describes development of an automatic generator of 3D models for Blender. After a brief introduction to the Blender application some of existing scripts for 3D model creation (tested as part of theoretical research) are presented. After an analysis of features required from an automatic generator various types of swords were chosen as the suitable class of objects for the automatic creation. Next part of this thesis presents basic terms in this field, namely the nomenclature of the swords and other related weapons. Their most important properties are listed and explained. The chapter also provides a simplified abstract of the sword typology. Following chapters describe main ideas of creating the model, explain principles and mathematics behind used algorithms and procedures and provide information about realization of the GUI as well as testing and optimisation. As the text explains these issues, sugestions about possible future enhancements are made. Document is concluded with an evaluation of all achievements and argues impact of this work on its field.

Trilobites were a successful group of marine arthropods that thrived for much of the Palaeozoic Era. Akey innovation of trilobites involved the capacity to enrol their biomineralized exoskeletons for protection; a trait that appeared early in their evolutionary history during the Cambrian Period. However, even though there has been much interest in the description of enrolment-related anatomical adaptations, and different styles of enrolment, the kinematic aspect has just begun to be explored—not just as means of understanding trilobite movements, but also as a potential driver of evolution. In this study, with the use of the open-source software Blender, four models of three different species of Ordovician trilobites(two phacopines and two illaenines) were constructed and later submitted to enrolling simulations. Different angle metrics were measured and then compared between simulations to analyse the performance of the developed methodology, and between the different models to reveal potential evolutionary or developmental trends. Results suggest a clear relationship between the kind of higher-level taxonomic groups analysed and the angle arrangement of the trunk elements in an enrolled position. In contrast, the same was not as evident when comparing the measured angle metrics against the total number of segments. Illaenina models consistently resulted in lower angles between segments required to enrol, suggesting a more efficient enclosure of the body favoured by bigger and more spherical proportions of cephalon and pygidium. This study contributes to a growing field in palaeontology and provides new, detailed enrolment data for several trilobite species. In future, such data should be synthesized with those from previous and future studies to analyse the evolutionary and developmental history of trilobites from this novel perspective.

Eine Zusammenstellung der Blender-Quellfiles, welche zum UNIX-STammtisch "Character-Animation mit Blender" gezeigt wurden.

A method to quickly produce large sets ofdifferent textures of insects (bugs) wasimplemented. The textures were procedurallygenerated in a 3D-modelling tool and thenrendered for use in a 2D-motion-based game.The resulting textures were evaluatedregarding quality and distinguishability. Noparticular features were found to be favouredover others, except some features being betterthan no features at all. Distinguishing betweengenerated textures was found to be a trivial taskfor most users.

Detta arbete har haft som syfte att undersöka huruvida Blenders (1994) tilläggsverktyg Rigify kan användas för att frigöra arbetskraft inom spelindustrin, då det är ett verktyg för riggning av 3D-karaktärer. Forskare har tidigare gjort försök till att skapa algoritmer som kan hjälpa till inom 3D-animation genom att ha skapat självgenererade riggar (Baran &amp; Popovic 2007; Pan et al 2009; Bharaj et al 2012), samt riggar som också skinnas, eller knyts, direkt till karaktären (Le &amp; Deng 2014). Undersökningen i detta arbete bestod utav en 3D-karaktär som riggas med hjälp av Rigify, som därefter utvärderades av anställda inom spelindustrin. Deltagare fick sedan dela sina åsikter om karaktärens rigg under en intervju. Resultatet visade på att Rigify kan skapa avancerade riggar, dock med mindre brister i genereringen. Riggen upplevdes som överflödig med de många funktioner som lades till. Rigify framstår som ett bra verktyg för mindre nystartade spelstudior, fritidsutvecklare, samt yrkesmässiga egenutvecklare. / <p>Det finns övrigt digitalt material (t.ex. film-, bild- eller ljudfiler) eller modeller/artefakter tillhörande examensarbetet som ska skickas till arkivet.</p>

The aim of the thesis is to help people visualize architectural designs more effectively. By using their smartphone cameras, users can superimpose digital information in a real-world environment. The study explores whether Augmented Reality (AR) can be practical in the field of building design, showcasing benefits such as saving costs and making the implementation process simpler. This research is noteworthy for its originality, as it focuses on the unexplored potential of AR in this particular area. The results emphasize how AR has the ability to make blueprint visualization more lifelike and dynamic, thereby enhancing the overall user experience.