Knowing the semantic category of a robot's current position not only facilitates the robot's navigation, but also greatly improves its ability to serve human needs and to interpret the scene. Visual Place Categorization (VPC) is addressed in this dissertation, which refers to the problem of predicting the semantic category of a place using visual information collected from an autonomous robot platform. Census Transform (CT) histogram and Histogram Intersection Kernel (HIK) based visual codebooks are proposed to represent an image. CT histogram encodes the stable spatial structure of an image that reflects the functionality of a location. It is suitable for categorizing places and has shown better performance than commonly used descriptors such as SIFT or Gist in the VPC task. HIK has been shown to work better than the Euclidean distance in classifying histograms. We extend it in an unsupervised manner to generate visual codebooks for the CT histogram descriptor. HIK codebooks help CT histogram to deal with the huge variations in VPC and improve system accuracy. A computational method is also proposed to generate HIK codebooks in an efficient way. The first significant VPC dataset in home environments is collected and is made publicly available, which is also used to evaluate the VPC system based on the proposed techniques. The VPC system achieves promising results for this challenging problem, especially for important categories such as bedroom, bathroom, and kitchen. The proposed techniques achieved higher accuracies than competing descriptors and visual codebook generation methods.

[pt] A segmentação de documentos HTML tem sido essencial para as tarefas de extração de informações, como mostram vários estudos na área. Nesta dissertação investigamos a relação entre o documento HTML e sua representação visual, mostrando como esta ligação ajuda na abordagem estrutural para a identificação de segmentos. Também investigamos como utilizar algoritmos de distância de edição em árvores para encontrar padrões na árvore DOM, tornando possível resolver duas tarefas de identificação de segmentos. A primeira tarefa é a identificação de tabelas genuínas, aonde foi obtido 90,40% de F1 utilizando o corpus fornecido por (Wang e Hu, 2002). Mostramos através de um estudo experimental que este resultado é competitivo com os melhores resultados da área. A segunda tarefa que consideramos é a identificação de listas de produtos em sites de comércio eletrônico, nessa obtivemos 94,95% de F1 utilizando um corpus com 1114 documentos HTML, criado a partir de 8 sites. Concluímos que os algoritmos de similaridade estrutural ajudam na resolução de ambas às tarefas e acreditamos que possam ajudar na identificação de outros tipos de segmentos. / [en] The segmentation of HTML documents has been essential to information extraction tasks, as showed by several works in this area. This paper studies the link between an HTML document and its visual representation to show how it helps segments identification using a structural approach. For this, we investigate how tree edit distance algorithms can find structural similarities in a DOM tree, using two tasks to execute our experiments. The first one is the identification of genuine tables where we obtained a 90.40% F1 score using the corpus provided by (Wang e Hu, 2002). We show through an experimental study that this result is competitive with the best results in the area. The second task studied is the identification of product listings in e-commerce sites. Here we get a 94.95% F1 score using a corpus with 1114 HTML documents from 8 distinct sites. We conclude that algorithms to calculate trees similarity provide competitive results for both tasks, making them also good candidates to identify other types of segments.

Domestic violence against women (DVAW) is a global pandemic that affects approximately one in three women living in the United Kingdom. One of the biggest challenges in combating and preventing DVAW is the underreporting of incidences to law-enforcement by victims and the collective silence of bystanders who are aware of the abuse but choose to remain silent (i.e. social silence). This degree project compares evidence regarding social silence and underreporting of DVAW and how DVAW is represented in British awareness campaigns to answer the research question: What gaps exist between evidence available in respect of contributing factors to social silence and underreporting of DVAW and how key players in the space of DVAW prevention in the United Kingdom (UK) represent DVAW in public awareness campaigns aimed at addressing these issues? A literature review served to establish state of the art evidence and was followed by a Foucauldian discourse analysis of selected visual media texts from awareness raising campaigns published by British key players (i.e. NGOs and government agencies) in the area of DVAW. The analysis was conducted in three stages: 1) relevant discourse fragments were identified according to strict sampling criteria, 2) texts were analysed with a step-by-step approach, in order to identify key themes and a typical sub-sample of discourse fragments, and 3) an in-depth analysis of two typical campaign texts was conducted. The analysis revealed that a range of misalignments and gaps exist. DVAW is represented in isolation as an issue of individuals rather than society. Some of the most problematic attitudes contributing to social silence and underreporting of DVAW such as victim blaming remain largely unaddressed. Victims of DVAW are represented in isolation and the responsibility to act and stop the abuse is often placed on the them. Perpetrators of DVAW in particular, but also men in general, are largely excluded from the discourse. Instead of encouraging victims and building their confidence, a bleak picture of isolation and fear is painted in campaign texts. Some of the discourse fragments included in the analysis appear to perpetuate the very misconceptions and stereotypes they are trying to address. There is ample opportunity for British key players in the space of DVAW to take a leading role in challenging the current discourse and assume their role of influencer in the fight to break social silence and increase the reporting of DVAW.

Based on fieldwork with human rights organizations in New York City and Belize, Central America, this dissertation explores--through the prism of ethics--how non-governmental organizations (NGOs) represent violence against indigenous women--in text, image, and action--as human rights "evidence." By ethics I mean the deliberate use of morals, stated or unstated, in the representation of human rights abuses. In New York, my research focuses on the production, launch, and circulation of a United Nations shadow report on violence against indigenous women. In Belize, I contextualize indigenous women's experiences of gender violence within an indigenous movement to obtain collective land rights, a national women's movement, and national rhetoric on culture and gender. In both locales, I consider and compare: 1) how the "ethical" stance of NGOs shapes human rights activism; 2) how NGOs create visual and discursive "evidence" to represent violence and indigenous women's experiences; and 3) very real neoliberal state repression that immobilizes social movements for human rights and social justice. My concern is with the ways social movement NGOs struggle to maintain their feminist and social justice objectives as they interface with the demands of a transnational human rights system, and the strategies they use as they suffer from vilification, marginalization or mainstreaming, and lack of resources. Far from protective, human rights claims, explored here as "evidence," often obscure both social inequalities and the response of state-level policies to these inequalities, especially for marginalized women. / Anthropology

Through an examination of Joseph Roth’s reportage and fiction published between 1923 and 1932, this thesis seeks to provide a systematic analysis of a particular aspect of the author’s literary style, namely his use of sharply focused visual representations, which are termed Heuristic Visuals. Close textual analysis, supplemented by insights from reader-response theory, psychology, psycholinguistics and sociology illuminate the function of these visual representations. The thesis also seeks to discover whether there are significant differences and correspondences in the use of visual representations between the reportage and fiction genres. Roth believed that writers should be engagiert, and that the truth could only be arrived at through close observation of reality, not subordinated to theory. The research analyses the techniques by which Roth challenges his readers and encourages them to discover the truth for themselves. Three basic variants of Heuristic Visuals are identified, and their use in different contexts, including that of dialectical presentations, is explored. There is evidence of the use of different variants of Heuristic Visuals according to the respective rhetorical demands of particular thematic issues. It has also been possible to establish synchronic correspondences between the different genres, and diachronic correspondences within genres. Although there are examples within the reportage where the entire article is based on an Heuristic Visual, the use of Heuristic Visuals cannot be seen as a key organizing principle in Roth’s work as a whole. As his mastery of the technique reaches its highest point in the early 1930s, Heuristic Visuals are often incorporated into the reconstruction of a complete sensory experience. Analysis of Roth’s heuristic use of visual representations has led to important insights, including a reinterpretation of the endings of Roth’s two most famous novels: Hiob and Radetzkymarsch.

L'objectif de cette thèse est d'améliorer les représentations des images et des vidéos dans le but d'obtenir une reconnaissance visuelle accrue, tant pour des entités spécifiques que pour des catégories plus génériques. Les contributions de cette thèse portent, pour l'essentiel, sur des méthodes de description du contenu visuel. Nous proposons des méthodes pour la recherche d'image par le contenu ou par des requêtes textuelles, ainsi que des méthodes pour la reconnaissance et la localisation d'action dans des vidéos. En recherche d'image, les contributions se fondent sur des méthodes `a base de plongements de Hamming. Tout d'abord, une méthode de comparaison asymétrique vecteur-`a-code est proposée pour améliorer la méthode originale, symétrique et utilisant une comparaison code-`a-code. Une méthode de classification fondée sur l'appariement de descripteurs locaux est ensuite proposée. Elle s'appuie sur une classification opérée dans un espace de similarités associées au plongement de Hamming. En reconnaissance d'action, les contributions portent essentiellement sur des meilleures manières d'exploiter et de représenter le mouvement. Finalement, une méthode de localisation est proposée. Elle utilise une partition de la vidéo en super-voxels, qui permet d'effectuer un échantillonnage 2D+t de suites de boîtes englobantes autour de zones spatio-temporelles d'intérêt. Elle s'appuie en particulier sur un critère de similarité associé au mouvement. Toutes les méthodes proposées sont évaluées sur des jeux de données publics. Ces expériences montrent que les méthodes proposées dans cette thèse améliorent l'état de l'art au moment de leur publication. / The subject of this thesis is about image and video representations for visual recognition. This thesis first focuses on image search, both for image and textual queries, and then considers the classification and the localization of actions in videos. In image retrieval, images similar to the query image are retrieved from a large dataset. On this front, we propose an asymmetric version of the Hamming Embedding method, where the comparison of query and database descriptors relies on a vector-to-binary code comparison. For image classification, where the task is to identify if an image contains any instance of the queried category, we propose a novel approach based on a match kernel between images, more specifically based on Hamming Embedding similarity. We also present an effective variant of the SIFT descriptor, which leads to a better classification accuracy. Action classification is improved by several methods to better employ the motion inherent to videos. This is done by dominant motion compensation, and by introducing a novel descriptor based on kinematic features of the visual flow. The last contribution is devoted to action localization, whose objective is to determine where and when the action of interest appears in the video. A selective sampling strategy produces 2D+t sequences of bounding boxes, which drastically reduces the candidate locations. The method advantageously exploits a criterion that takes in account how motion related to actions deviates from the background motion. We thoroughly evaluated all the proposed methods on real world images and videos from challenging benchmarks. Our methods outperform the previously published related state of the art and remains competitive with the subsequently proposed methods.

HIV/AIDS research has proven crucial in an effort to prevent and manage this epidemic. However, there is little research being done in an attempt to understand the internal worlds of those living with HIV/AIDS. The purpose of this research was to begin to explore the relationship that exists between the person living with HIV/AIDS and the virus, as an internal object, inside them. This study focused on 6 women who were on Anti-Retroviral Medication (ARVs), and who have been diagnosed for at least one year. The participants’ mental representations of the virus as an object inside them was explored, as well as how they experienced and viewed the triangular relationship that exists between themselves, the HI Virus, and the ARVs. This exploratory research utilised a qualitative framework in order to understand and explore these relationships and perceptions, with psychoanalytic theory being used a lens through which to view the data that emerged. In depth semi-structured interviews were conducted with the participants, and the corpus of data was analyzed using a thematic content analysis. In addition, the participants were asked to draw the virus inside their bodies. These were analyzed using a technique devised by Paola Luzzatto (1987) in a study exploring the internal world of drug-abusers. For the purpose of this study, a variation of the same art therapy technique was used in that the participants were asked the ‘draw the virus in their bodies’. Whilst the drawings allowed for insight into the internal worlds of the participants, the drawings were also used as a point of departure. For most of the women, HIV was drawn using a red crayon, whilst the ARVs were drawn in either yellow or green. As depicted in the drawings, post diagnosis the HIV/red seemed to cover most of the body, but later when the ARVs/green was added, more of a balance was achieved. Results show that for these women, HIV was often perceived as dangerous and criminal, whilst the ARVs were often associated with security. From the perspective of Kleinian theory, the perception of the HIV and the ARVs seemed to be dependent upon the position from where they were functioning: either a paranoid-schizoid or a depressive position.

Background - Welcome to the 2010’s and its visual information culture informing about the century-old ideal of thinness, by sending (almost) exclusively thin bodies down the runway of Paris Fashion Week (PFW). This, while at the same time the academia and mass media are repeatedly articulating the necessity of a more ethical aesthetic within the fashion industry, while the argument that “thinness sells” lack empirical support, while measures are taken around Europe to discourage media bias and encourage healthy bodies walking the, and while the body positivity movement – applauding the diversity of body types – is on the rise in the Western society. Problem - How can it be explained that PFW, as an event organized to communicate the latest trends, and as an institution of a concept that thrives on change, is being stuck in an inertia of thin models? The purpose of this qualitative study is to explore the explanations for the idealization of thin bodies in regard to female runway models, to be found in visual representations of PFW. Examining the paradox of the idealization of thin bodies has the potential to reveal the measures necessary to establish practices of ethical representation in the realm of PFW. Method - The search for explanations to the inertia of the thin ideal on the runways of PFW is based upon a qualitative research design and a method of semi-structured interviews. The sample consisted of fourteen people involved in the event of PFW. It is framed within three theoretical perspectives: the trickle-down, trickle-across and trickle-up theories – associated with authors such as Simmel (1957), King (1963) and Blumer (1969a; 1969b); a theoretical view of the relation between fashion, clothes and the body provided by Barnard (2014) and, lastly, the Organizational Ecology Theory presented by Hannan and Freeman (1977; 1989). Findings - The findings suggest that the inertia of the thin ideal on the runways of PFW is explained by the thin body not only being part of a long history and tradition of creating and showing fashion, but also being dressed in aesthetic value. At the same time, the explanations to the idealization, or inertia, of thin bodies on PFW are found to be multiple as well as paradoxical, a conclusion serving as an explanation in itself. Conclusion – It is difficult to point out the explanation to the inertia of the thin ideal – both in terms of factors and actors – not at least due to the explanations stemming from both social and cultural as well as economic aspects of fashion. As a consequence, every actor part the fashion industry need to understand the effects as well as the extent of idealization of thin bodies, and start to realize that public health is more important than profits.

Puzzles have always been a central part of games. One of the simplest manifestations of a puzzle in games is an obstruction that prevents the player from progressing and an object that removes the obstruction. The system of unlocking new areas with keys is used in order to provide interesting level design and help build narrative in games. This examination intends to provide knowledge and insight into how this type of system is commonly portrayed visually and thus provide a basis for designers to build upon when creating similar challenges, providing a greater awareness of the design norms that are commonly applied. In the majority of the cases that have been examined during this study, text-based information is used to convey the majority of the essential information given to the players regarding the relationship of the key and the lock. In many cases the graphics serve only to highlight important features of the objects and help the player to quickly tell the different items apart in the inventory.

No description available.