Ett nytt kommunhus för Nynäshamn stad, belägen i Stockholms län sydspets, vilket innehåller aktivitetsbaserade kontor, ett konstgalleri, kafé och en plenisal. Byggnaden ligger mellan två olika men kompletterande stadsrum vilket byggnaden bemöter med två olika "ansikten", allt medan den fortfarande uppfattas som samma byggnad. Denna idé, vilken kan liknas vid idén om Janusansiktet, har varit min huvudsakliga ingång i det här projektet. Utvändigt både återspeglar ansiktena de stadsrum som äger rum framför och är aktiva i formandet av dess karaktärer. Sammanflätningen mellan de båda sidorna sker istället inuti byggnaden där övergången ömsom är flytande och ömsom skarp med visuella snarare än fysiska kopplingar. / A new municipal building for the city of Nynäshamn, located in the most southern part of Greater Stockholm, including acitivity based offices, an art gallery, café and council chamber. The building lies inbetween two different but complementary public spaces which it adresses with two different "faces" while still being read as one and the same building. This idea, reminiscent of the image of the roman god Janus, has been my main focus through this project. Outwards, the faces both reflect and take an active roll in forming the character of these public spaces while the interweaving takes places inside the building where they sometimes are visually, although not physically connected, and sometimes closer to fade into one another.

[pt] A quantidade de vídeos gravados no mundo cresce muito, não somente devido aos interesses e hábitos humanos em relação a esse tipo de mídia, mas também pela diversidade de dispositivos utilizados para criação de vídeos. No entanto, faltam informações sobre conteúdos em vídeo porque a geração de metadados é complexa e requer muito tempo para ser executado por humanos. Do ponto de vista da tecnologia, não é fácil superar os obstáculos relacionados à grande quantidade e diversidade de frames de vídeo. O trabalho propõe um sistema automatizado de reconhecimento facial para detectar personagens em vídeos. Ele foi desenvolvido para reconhecer personagens, a fim de aumentar os metadados de vídeo. Ele combina técnicas padrão de visão computacional para melhorar a precisão, processando os dados de saída dos modelos existentes de maneira complementar. O modelo teve um desempenho satisfatório usando um conjunto de dados da vida real de uma grande empresa de mídia. / [en] The amount of recorded video in the world is increasing a lot due not only to the humans interests and habits regarding this kind of media, but also the diversity of devices used to create them. However, there is a lack of information about video content because generating video meta-data is complex. It demands too much time to be performed by humans, and from the technology perspective, it is not easy to overcome obstacles regarding the huge amount and diversity of video frames. In this work we propose an automated face recognition system to detect and recognize humans within videos. It was developed to recognize characters,in order to increase video meta-data. It combines standard computer vision techniques to improved accuracy by processing existing models output data in a complementary manner. We evaluated the performance of the system in a real data set from a large media company.

[pt] Esta dissertação propõe e avalia estratégias baseadas nos Stacked Supervised Auto-encoders (SSAE) para representação de imagens faciais em aplicações de vídeo vigilância. O estudo foca na identificação de faces a partir de uma amostra por pessoa na galeria (single sample per person - SSPP). Variações em termos de pose, expressão facial, iluminação e oclusão são abordadas de duas formas. Primeiro, o SSAE extrai atributos das imagens de faces que são robustos contra tais variações. Segundo, exploram-se as múltiplas amostras que podem ser coletadas nas sequências de vídeo de uma pessoa (multiple samples per person probe - MSPPP). Os métodos propostos foram avaliados e comparados usando os bancos de vídeos Honda/UCSD e VIDTIMIT. Adicionalmente, foi estudada a influência de parâmetros relacionados com a arquitetura do SSAE utilizando o banco de imagens estáticas Extended Yale B. Os resultados demonstraram que as estratégias que exploram as MSPPP em combinação com o SSAE podem superar o desempenho de outros métodos SSPP, como os Padrões Binários Locais (LBP), para reconhecimento de faces em vídeos. / [en] This work proposes and evaluates strategies based on Stacked Supervised Auto-encoders (SSAE) for face representation in video surveillance applications. The study focuses on the identification task with a single sample per person (SSPP) in the gallery. Variations in terms of pose, facial expression, illumination and occlusion are approached in two ways. First, the SSAE extracts features from face images, which are robust to such variations. Second, multiple samples per persons probes (MSPPP) that can be extracted from video sequences are exploited to improve recognition accuracy. The proposed methods were compared upon Honda/UCSD and VIDTIMIT video datasets. Additionally, the influence of the parameters related to SSAE architecture was studied using the Extended Yale B dataset. The experimental results demonstrated that strategies combining SSAE and MSPPP are able to outperform other SSPP methods, such as local binary patterns, in face recognition from video.

The structural and functional differences observed in the brains of adults with synaesthesia is thought to arise, at least in part, from less-than-normal neural pruning of the exuberant connections present within and among sensory cortical areas in infancy (reviewed in Maurer, Gibson, & Spector, 2013). This hypothesis is supported by previous work that has demonstrated that synaesthetes are superior at processing foreign speech sounds and inverted faces (Maurer et al., in prep). The present study investigated a link between spatial frequency and face processing in adults with synaesthesia by testing synaesthetes and non-synaesthetes on their ability to discriminate upright and inverted faces filtered at high and low spatial frequencies. As predicted, synaesthetes (n=20) were significantly more accurate than non-synaesthetes (n=20) at discriminating among inverted full spectrum faces (p=0.0235), with no differences in upright faces, replicating previous findings that support the hypothesis that synaesthetes undergo less perceptual attunement (Ghloum et al., 2013). Unexpectedly, synaesthetes were faster at responding across all face conditions. Faster reaction times with no sacrifice to accuracy suggest that synaesthetes may be processing faces more efficiently. In addition, no significant differences in accuracy were observed for high and low filtered faces at any orientation between synaesthetes and non-synaesthetes. Future studies could further explore the basis of synaesthete’s face processing advantages by using eye movements and a narrow-band noise-masking paradigm. / Thesis / Master of Science (MSc)

No description available.

No description available.

No description available.

No description available.

[pt] O problema tratado nesta dissertação consiste em modelar reservatórios fluviais utilizando modelos geoestatísticos baseados em objetos. Este trabalho propõe um modelo baseado em objetos para sistemas fluviais que simula, conforme dados observados, duas das principais faces associadas aos reservatórios fluviais: o canal e o espraiamento de crevasse. O modelo proposto realiza uma simulação com condicionantes baseado na técnica Simulated Annealing. / [en] The problem addressed in this work is to model fluvial reservoirs using geostatistical model based on objects. This work aims to create an object-based model for fluvial systems which simulates, given some observed data, two of the principal facies associated to the fluvial reservoirs: the channel sand and crevasse splay sand. Also, it applies a simulation with constraints based on the Simulated Annealing approach.

Sandwich panels comprise a thick, light-weight plastic foam such as polyurethane, polystyrene or mineral wool sandwiched between two relatively thin steel faces. One or both steel faces may be flat, lightly profiled or fully profiled. Until recently sandwich panel construction in Australia has been limited to cold-storage buildings due to the lack of design methods and data. However, in recent times, its use has increased significantly due to their widespread structural applications in building systems. Structural sandwich panels generally used in Australia comprise of polystyrene foam core and thinner (0.42 mm) and high strength (minimum yield stress of 550 MPa and reduced ductility) steel faces bonded together using separate adhesives. Sandwich panels exhibit various types of buckling behaviour depending on the types of faces used. Three types of buckling modes can be observed which are local buckling of plate elements of fully profiled faces, flexural wrinkling of flat and lightly profiled faces and mixed mode buckling of lightly profiled faces due to the interaction of local buckling and flexural wrinkling. To study the structural performance and develop appropriate design rules for sandwich panels, all these buckling failure modes have to be investigated thoroughly. A well established analytical solution exists for the design of flat faced sandwich panels, however, the design solutions for local buckling of fully profiled sandwich panels and mixed mode buckling of lightly profiled sandwich panels are not adequate. Therefore an extensive research program was undertaken to investigate the local buckling behaviour of fully profiled sandwich panels and the mixed mode buckling behaviour of lightly profiled sandwich panels. The first phase of this research was based on a series of laboratory experiments and numerical analyses of 50 foam-supported steel plate elements to study the local buckling behaviour of fully profiled sandwich panels made of thin steel faces and polystyrene foam core covering a wide range of b/t ratios. The current European design standard recommends the use of a modified effective width approach to include the local buckling effects in design. However, the experimental and numerical results revealed that this design method can predict reasonable strength for sandwich panels with low b/t ratios (< 100), but it predicts unconservative strengths for panels with slender plates (high b/t ratios). The use of sandwich panels with high b/t ratios is very common in practical design due to the increasing use of thinner and high strength steel plates. Therefore an improved design rule was developed based on the numerical results that can be used for fully profiled sandwich panels with any practical b/t ratio up to 600. The new improved design rule was validated using six full-scale experiments of profiled sandwich panels and hence can be used to develop safe and economical design solutions. The second phase of this research was based on a series of laboratory experiments and numerical analyses on lightly profiled sandwich panels to study the mixed mode buckling behaviour due to the interaction of local buckling and flexural wrinkling. The current wrinkling formula, which is a simple modification of the methods utilized for flat panels, does not consider the possible interaction between these two buckling modes. As the rib depth and width of flat plates between the ribs increase, flat plate buckling can occur leading to the failure of the entire panel due to the interaction between local buckling and wrinkling modes. Experimental and numerical results from this research confirmed that the current wrinkling formula for lightly profiled sandwich panels based on the elastic half-space method is inadequate in its present form. Hence an improved equation was developed based on validated finite element analysis results to take into account the interaction of the two buckling modes. This new interactive buckling formula can be used to determine the true value of interactive buckling stress for safe and economical design of lightly profiled sandwich panels. This thesis presents the details of experimental investigations and finite element analyses conducted to study the local buckling behaviour of fully profiled sandwich panels and the mixed mode buckling behaviour of lightly profiled sandwich panels. It includes development and validation of suitable numerical and experimental models, and the results. Current design rules are reviewed and new improved design rules are developed based on the results from this research.