Common‒sense knowledge is a collection of non‒expert and agreed‒upon facts and information about the world shared among most people past early childhood based on their experiences. It includes uses of objects, their properties, parts and materials, their locations, spatial arrangements among them; location and duration of events; arguments, preconditions and effects of actions; urges and emotions of people, etc. In creating 3D worlds and especially text‒ to‒scene and text‒to‒animation systems, this knowledge is essential to eliminate the tedious and low‒level tasks, thus allowing users to focus on their creativity and imagination. We address tasks related to five categories of common‒sense knowledge that is required by such systems including: (1) spatial role labeling to automatically identify and annotate a set of spatial signals within a scene description in natural language; (2) grounding spatial relations to automatically position an object in 3D world; (3) inferring spatial relations to extract symbolic spatial relation between objects to answer questions regarding a 3D world; (4) recommending objects and their relative spatial relations given a recent manipulated object to auto‒complete a scene design; (5) learning physical attributes (e.g., size, weight, and speed) of objects and their corresponding distribution. We approach these tasks by using deep learning and probabilistic graphical models and exploit existing datasets and web content to learn the corresponding common‒sense knowledge.

3D geological modelling is becoming an effective tool for communication and development of geological understanding. This is due to increased computer performance and availability of improved geological modelling software. 3D geological modelling technology has reached the stage where it can be implemented in regionally extensive and geologically complex settings, with the ability to achieve geological insight beyond what could otherwise have been gained through 2D investigations alone. Insight includes better constrained fault and horizon topologies, refined fold geometries, improved understanding of tectonic processes, and characterization of deformational events. By integrating field observations, aeromagnetic maps, and 3D modelling techniques in the northern Labrador Trough, a regionally extensive and structurally complex geological environment, regional faults geometries and topological relationships were refined. Additionally, a new fault, the Ujaralialuk Fault, and two shear zones were interpreted. During modelling, several challenges were identified, including higher computational costs for regionally extensive models, sparse 3D constraints, algorithmic limitations related to complex geometries, and the large investment of time and effort required to produce a single model solution. A benefit of this investigation is that new insight was also gained for a greenfields region which may assist future exploration efforts. Developing 3D models in challenging environments allows for better definition of future workflow requirements, algorithm enhancements, and knowledge integration. These are needed to achieve a geologically reasonable modelling standard and gain insight for poorly constrained geological settings.

Orientadores: Marco Antonio Garcia de Carvalho, Marco Aurelio Takita / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Tecnologia / Made available in DSpace on 2018-08-26T15:21:12Z (GMT). No. of bitstreams: 1 Palma_EloaCarolinaNavaCardoso_M.pdf: 2004126 bytes, checksum: 618ad5da7f91e98f63bfdc0a41812387 (MD5) Previous issue date: 2014 / Resumo: O surgimento dos sequenciadores automáticos de DNA por volta dos anos 1990 fez com que a quantidade de sequências e informações a serem analisadas aumentasse exponencialmente. Sendo assim, os recursos computacionais são cada vez mais exigidos em diversas tarefas de armazenamento desses dados. Para atender essas e outras necessidades, surgiu a Bioinformática, que engloba diversas áreas como Ciência da Computação, Matemática, Estatística, e a Biologia Molecular. Neste trabalho, desenvolvemos um software denominado Biodonut, que consiste em manipular plasmídeos utilizando visualização 2D e 3D. Biodonut é um produto de software de plataforma web, open source para que os pesquisadores possam manipular plasmídeos in silico, visualizando em formato 2D e 3D e possibilitando uma análise diferenciada das áreas julgadas importantes. A base de dados utilizada no software é o GenBank, por ser mundialmente conhecida e respeitada. Uma comparação com dois importantes softwares da área foi efetuada e é apresentada neste trabalho. Os resultados apontam que usar Biodonut é vantajoso em relação os outros dois softwares / Abstract: The advent of automated DNA sequencers around 1990s led to the amount of information sequences to be analyzed and increased exponentially. Thus, computing resources are increasingly demanded in various tasks such data storage. To meet these and other needs arose Bioinformatics, encompassing diverse areas as Computer Science, Mathematics, Statistics, and Molecular Biology. In this work, we developed a software called Biodonut, which involves manipulating plasmids using 2D and 3D visualization. Biodonut is a product of web software platform, open source so that researchers can manipulate plasmids in silico, visualizing 2D and 3D format, enabling a differentiated analysis of the areas deemed important. The database software is used in GenBank, to be globally known and respected. A comparison with two important software of the area was conducted and is presented in this paper. The results show that using Biodonut is advantageous over the other two softwares / Mestrado / Tecnologia e Inovação / Mestra em Tecnologia

This thesis deals with design, construction and software of intelligent warehouse for testbed Industry 4.0. The first part is dedicated to the issue of warehousing, storage, Industry 4.0 and the description of the testbed Barman. The second part deals with the development of the cell storage. Firstly, there is a description of the design and construction of the cell, followed by a description of the sensors and control equipment used. The following part deals with the software of the cell and finally the goals of the thesis are evaluated.

Les modélisations 3D de ville se multiplient à travers le monde et deviennent aujourd’hui accessibles grâce à la volonté des communes de les proposer librement. Il est ainsi aujourd’hui possible d’accéder à plusieurs milliers de kilomètres carrés de territoires urbains modélisés en 3D. Nous présentons dans cette thèse un ensemble de méthodes permettant d’enrichir un modèle virtuel 3D de ville, de l’organiser afin de faciliter son utilisation, puis de l’analyser en détectant les changements entre deux millésimes ou en mesurant son ambiance visuelle selon les besoins de l’utilisateur. Nous proposons dans un premier temps un algorithme permettant de compléter un modèle 3D en y adjoignant une définition sémantique grâce au cadastre, ainsi qu’une méthode de calcul de la canopée végétale 3D par croisement de données. Puis nous proposons des stratégies d’organisation du modèle 3D de la ville selon des critères géométriques et sémantiques afin de faciliter son parcours. Nous présentons ensuite des méthodes comparant deux millésimes de données représentant un même territoire dans le but de produire un unique modèle temporel contenant un ensemble de versions. Enfin, nous cherchons à mesurer l’ambiance visuelle de la ville via l’analyse de la composition d’un champ de vision ainsi que de l’ensoleillement et des ombres portées présentes sur un territoire. Ces mesures peuvent être liées à des données externes afin de proposer de multiples interprétations en fonction des besoins de l’utilisateur.Tous ces travaux se font dans un contexte d’interopérabilité et de généricité puisqu’ils ont pour objectif de pouvoir être utilisés avec des jeux de données provenant du monde entier. Nous basons donc nos méthodes sur l’utilisation de standards internationaux aussi bien pour les données en entrée que pour les résultats en sortie. Il est ainsi possible, dans le cadre d’une approche voulue dans un contexte de pluralité scientifique au sein du LabEx IMU, de mettre à disposition de la communauté les résultats mais aussi nos algorithmes proposés au sein d’un logiciel développé en open source. / 3D virtual models of cities are multiplying throughout the world and now become accessible thanks to the will of the communes to propose them freely. Today, it is possible to access several thousand square kilometres of urban territories modelled in 3D.We present in this thesis a set of methods to enrich a 3D virtual city model, organize it to facilitate its use and then analyse it by detecting changes between two vintages or by measuring its visual atmosphere according to the user’s needs. We first propose an algorithm to complete a 3D model by adding a semantic definition thanks to the cadastre, and a method to compute the 3D plant canopy by crossing data. Then we propose strategies to organize the 3D model of the city according to geometric and semantic criteria in order to facilitate its browsing. We then present methods comparing two vintages of data representing the same territory in order to produce a single temporal model containing a set of versions. Finally, we try to measure the visual atmosphere of the city by analysing the composition of a field of vision as well as the amount of sunshine and the shadows on a territory. These measurements can be linked to external data in order to propose multiple interpretations according to the user’s needs.All this work is done in a context of interoperability and genericity since it aims to be used with datasets from all over the world. We therefore base our methods on the use of international standards for both input and output data. It is thus possible, in a context of scientific plurality within the LabEx IMU, to make available to the community the results but also our algorithms proposed within an open source developed software.

Tall buildings have been constructed in many cities over the world to accommodate the demographic and economic growth in urban areas. These structures, however, have been widely discussed concerning their potential negative impacts on urban life. By drawing on concepts of urban morphology and the city image, this thesis aims to examine the impact of tall buildings on predominantly low-rise built urban environments and cities. Through a quantitative case study approach applying geographic information systems this thesis attempts to identify (1) the shadow impacts significantly taller buildings have on surrounding areas, (2) the visual impact and the potential for a proposed tall building to act as a new landmark, and (3) the impact tall buildings have on the skyline of a city dominated by low-rise buildings and a historic city centre. The thesis analysed two proposed buildings in Västerås, Sweden and found that the shadow impact of the analysed structures is concentrated on the winter months. The visibility of the buildings concentrates on open spaces and certain visible corridors but is not widely spread over the city. Therefore, the possibilities of the new structures becoming a new landmark of the city are questionable. Furthermore, the proposed planned structures would be visible in the skyline but would not obstruct the views to the other landmarks of the city. The thesis and its results contribute to the discourse around tall buildings and the research on the city image by showing the importance of the analysis of the potential impact of tall buildings in urban planning processes.

This thesis compares two modeling methods, subdivision and booleans, which are used for the creation of hard surface assets in the gaming industry. The parameters that are assessed are the time taken to create the models with the respective method and the quality of the models. Quality refers to the presence of shading artifacts. In order to compare the methods, models of varying complexity have been created, the time of creation has been measured and experts in 3D modeling have assessed the quality. The results show that the boolean method is better suited for complex models, while both methods work well for simpler models. / Den här avhandlingen jämför två modelleringsmetoder, subdivision och booleans, som används för skapandet av hardsurface assets inom spelbranschen. Parametrarna som bedöms är den tid det tar att skapa modellerna med respektive metod och kvalitén på modellerna. Med kvalité avses förekomsten av shading artefakter. För att jämföra metoderna har modeller av varierande komplexitet skapats, skapandets tid har mätts och experter inom 3D-modellering har bedömt kvalitén. Resultatet visar att booleansmetoden är bättre lämpad för komplexa modeller medan båda metoder fungerar bra för simplare modeller.

Located on the island of Gotland, Sweden, the File Hajdar limestone quarry is an important locality for domestic cement production. With depositional settings generally related to shallow sea carbonate ramps, the lithologies encountered on File Hajdar range from reef-associated limestone units to more basinal deposition of marls. As the depositional setting is related to the CaCO3-content (carbonate), which is an important property to consider during cement production, geological investigation of the area through conceptual 3D modelling could be a very powerful tool to enhance the general knowledge of the area or for planning future prospecting campaigns. Since the quality of the 3D model is related to the amount and quality of data utilized, data acquisition and management are very important steps in the construction of a 3D model. Two main campaigns with drill core extractions were done in 1993 and 1999, respectively. Along with resistivity data collected from skyTEM-measurements and ground surveys, subsurface geological information is quite extensive in the area. Using overall data and a reclassification of the lithologies, three main areas associated with reef complex deposition have been identified on File Hajdar. Corresponding to elevated resistivity measurements as well as classification in drill cores, two reef complexes centered around the quarry, have been described as the central- and southeastern reef complexes. Based on high resistivity anomalies in skyTEM-data along with field interpretations, the emplacement of a third reef complex is also interpreted in the northwestern parts of the study area (northwestern reef complex). However, as overall data is lacking at the location of the northwestern reef complex, further investigation is necessary before the relationship between the central- and northwestern reef complex can be fully investigated. Due to the generally indistinct transition between clay-rich limestone and marl, the subsurface boundary to the marl is rather insinuated. However, corresponding to low resistivity anomalies, outcroppings of the surfacing marl unit are outlined at the base of the File Hajdar rise. Future utilization of the produced 3D model could include the implementation of additional data or as a foundation during the planning of further prospecting campaigns in the area. Corresponding to the locations of the central- and northwestern reef complexes, these two areas are suggested to be most suitable for further investigation. / Som en viktig komponent i cementproduktion har Gotland sedimentära berggrund haft en historisk anknytning till kalstensutvinning. File hajdar brottet är beläget ca 5 km väst om tätorten Slite, på nordöstra Gotland, och står i dagsläget helt för den lokala kalkstensutvinningen av Cementa AB. Kalkstensenheterna på File hajdar kan generellt kopplas till revbildning under den geologiska silurperioden. Men baserat på den interna relationen till de innersta revkropparna kan även underklassificeringar av kalkstensenheterna konstrueras. Eftersom relationen till reven även är kopplat till kalkstenens karbonathalt (CaCO3), vilket är en väsentlig beståndsdel att beakta under cementproduktion, är det viktigt att urskilja mellan de olika kalkstensenheter som påträffas på File hajdar. En sådan urskillning kan visuellt uppnås om en konceptuell 3D-modell över geologin på File hajdar brottet och dess omnejd konstrueras. Med data från borrkärnor, framtagna under två prospekteringsprojekt, samt från mätningar av berggrundens elektriska egenskaper kan man indirekt tolka geologin i området. Då det mer massiva revkropparna förväntas ha sämre ledningsförmåga relativt de mer lerberikade märgel- och kalkstenarna, kan man anknyta områden med hög resistivitet (oförmåga att leda ström) till revstrukturer. Från mätningar tolkas området norr- och söder om täkten inkludera större revstrukturer. Även nordväst om täkten kan man antyda att en större revstruktur är lokaliserad. Men eftersom den nordvästliga strukturen sakna större utsträckning av data behöver denna studeras närmare. Då övergången mellan den underliggande märgelstenen och File hajdars ytliga kalkstenen sker gradvis blir gränsen mellan dessa två enheter svår att placera ut med någon större säkerhet. Detta medför att den underliggande märgelgränsen bör betraktas som en generaliserad övergång i den slutgiltiga 3D-modellen. Den resulterade 3D-modellen ger en bra överblick över den generella geologin som påträffas i File hajdar området. Praktisk tillämpning av 3D-modellen kan exempelvis innefattar vidare tänktplanering eller fortsatt utvärdering av området genom implementering av mer data. Baserat på 3D-modellen framställs två förslag på lokaler lämpliga för vidare undersökningar. Motsvarande placeringarna av de norra- och nordvästra rekvkropparna tolkas dessa utgöra potentiella områden för kalkbrytning och i behov av ytterligare undersökning.

<p>A new approach for object replacement in 3D space is presented. Introducing a technique that replaces the older two dimensional (2D) based facial replacement method performed by compositing artist in motion picture productions and video commercial industry.</p><p>This method uses 4 digital video cameras filming an actor from 360 degrees, the cameras are placed with 90 degrees in between, the video footage acquired is then used to produce a 3D video texture consisting of video segments taken from different angles representing the object from 3D point of view.</p><p>The video texture is then applied to a 3D modelled head matching the geometry of the original object.</p><p>Offering the freedom of showing the object from any point of view from 3D space, which is not possible using the current two dimensional method where the actormust at all time face the camera.</p><p>The method is described in details with images showing every stage of the process.</p><p>Results are presented as still frames taken from the final video footage and as a video file demonstrating them.</p>

En aplicaciones de captura robotizada se ha hecho necesario el uso de información tridimensional de los objetos que son manipulados. Esta información puede obtenerse mediante dispositivos de adquisición 3D, tales como escáneres láser o cámaras de tiempo de vuelo, que proporcionan imágenes de rango de los objetos. En este trabajo de tesis se presenta un nuevo enfoque para encontrar, sin disponer de una estimación previa, la transformación rígida que produzca una alineación adecuada de las nubes de puntos obtenidas con esos dispositivos. El algoritmo realiza una búsqueda iterativa de correspondencias mediante la comparación de descriptores 2D en varios niveles de resolución utilizando para ello una medida de similitud específicamente diseñada para el descriptor propuesto en esta tesis. Este algoritmo de alineamiento se puede utilizar tanto para modelado 3D como para aplicaciones de manipulación de objetos en situaciones en las que los objetos estén parcialmente ocluidos o presenten simetrías. / In applications of robotic manipulation of objects, the use of three-dimensional information of objects being manipulated has been made necessary. This information can be obtained by 3D acquisition devices, such as laser scanners or cameras of flight time, providing range images of objects. This thesis presents a new approach to find, without having a previous estimate, the Euclidean transformation that produces a proper alignment of point clouds obtained with these devices. The algorithm performs an iterative search for correspondences by comparing 2D descriptors at various levels of resolution using a similarity measure specifically designed for the descriptor proposed in this thesis. This alignment algorithm can be used for both 3D modelling and robotic manipulation applications when objects are partially occluded or have symmetries.