The role of context in image annotation and recommendation

With the rise of smart phones, lifelogging devices (e.g. Google Glass) and popularity of image sharing websites (e.g. Flickr), users are capturing and sharing every aspect of their life online producing a wealth of visual content. Of these uploaded images, the majority are poorly annotated or exist in complete semantic isolation making the process of building retrieval systems difficult as one must firstly understand the meaning of an image in order to retrieve it. To alleviate this problem, many image sharing websites offer manual annotation tools which allow the user to “tag” their photos, however, these techniques are laborious and as a result have been poorly adopted; Sigurbjörnsson and van Zwol (2008) showed that 64% of images uploaded to Flickr are annotated with < 4 tags. Due to this, an entire body of research has focused on the automatic annotation of images (Hanbury, 2008; Smeulders et al., 2000; Zhang et al., 2012a) where one attempts to bridge the semantic gap between an image’s appearance and meaning e.g. the objects present. Despite two decades of research the semantic gap still largely exists and as a result automatic annotation models often offer unsatisfactory performance for industrial implementation. Further, these techniques can only annotate what they see, thus ignoring the “bigger picture” surrounding an image (e.g. its location, the event, the people present etc). Much work has therefore focused on building photo tag recommendation (PTR) methods which aid the user in the annotation process by suggesting tags related to those already present. These works have mainly focused on computing relationships between tags based on historical images e.g. that NY and timessquare co-exist in many images and are therefore highly correlated. However, tags are inherently noisy, sparse and ill-defined often resulting in poor PTR accuracy e.g. does NY refer to New York or New Year? This thesis proposes the exploitation of an image’s context which, unlike textual evidences, is always present, in order to alleviate this ambiguity in the tag recommendation process. Specifically we exploit the “what, who, where, when and how” of the image capture process in order to complement textual evidences in various photo tag recommendation and retrieval scenarios. In part II, we combine text, content-based (e.g. # of faces present) and contextual (e.g. day-of-the-week taken) signals for tag recommendation purposes, achieving up to a 75% improvement to precision@5 in comparison to a text-only TF-IDF baseline. We then consider external knowledge sources (i.e. Wikipedia & Twitter) as an alternative to (slower moving) Flickr in order to build recommendation models on, showing that similar accuracy could be achieved on these faster moving, yet entirely textual, datasets. In part II, we also highlight the merits of diversifying tag recommendation lists before discussing at length various problems with existing automatic image annotation and photo tag recommendation evaluation collections. In part III, we propose three new image retrieval scenarios, namely “visual event summarisation”, “image popularity prediction” and “lifelog summarisation”. In the first scenario, we attempt to produce a rank of relevant and diverse images for various news events by (i) removing irrelevant images such memes and visual duplicates (ii) before semantically clustering images based on the tweets in which they were originally posted. Using this approach, we were able to achieve over 50% precision for images in the top 5 ranks. In the second retrieval scenario, we show that by combining contextual and content-based features from images, we are able to predict if it will become “popular” (or not) with 74% accuracy, using an SVM classifier. Finally, in chapter 9 we employ blur detection and perceptual-hash clustering in order to remove noisy images from lifelogs, before combining visual and geo-temporal signals in order to capture a user’s “key moments” within their day. We believe that the results of this thesis show an important step towards building effective image retrieval models when there lacks sufficient textual content (i.e. a cold start).

Correlations among behavior, performance and environment in broiler breeders using multivariate analysis

Animal welfare issues have received much attention not only to supply farmed animal requirements, but also to ethical and cultural public concerns. Daily collected information, as well as the systematic follow-up of production stages, produces important statistical data for production assessment and control, as well as for improvement possibilities. In this scenario, this research study analyzed behavioral, production, and environmental data using Main Component Multivariable Analysis, which correlated observed behaviors, recorded using video cameras and electronic identification, with performance parameters of female broiler breeders. The aim was to start building a system to support decision-making in broiler breeder housing, based on bird behavioral parameters. Birds were housed in an environmental chamber, with three pens with different controlled environments. Bird sensitivity to environmental conditions were indicated by their behaviors, stressing the importance of behavioral observations for modern poultry management. A strong association between performance parameters and the behavior at the nest, suggesting that this behavior may be used to predict productivity. The behaviors of ruffling feathers, opening wings, preening, and at the drinker were negatively correlated with environmental temperature, suggesting that the increase of in the frequency of these behaviors indicate improvement of thermal welfare.

Machine learning techniques for implicit interaction using mobile sensors

Interactions in mobile devices normally happen in an explicit manner, which means that they are initiated by the users. Yet, users are typically unaware that they also interact implicitly with their devices. For instance, our hand pose changes naturally when we type text messages. Whilst the touchscreen captures finger touches, hand movements during this interaction however are unused. If this implicit hand movement is observed, it can be used as additional information to support or to enhance the users’ text entry experience. This thesis investigates how implicit sensing can be used to improve existing, standard interaction technique qualities. In particular, this thesis looks into enhancing front-of-device interaction through back-of-device and hand movement implicit sensing. We propose the investigation through machine learning techniques. We look into problems on how sensor data via implicit sensing can be used to predict a certain aspect of an interaction. For instance, one of the questions that this thesis attempts to answer is whether hand movement during a touch targeting task correlates with the touch position. This is a complex relationship to understand but can be best explained through machine learning. Using machine learning as a tool, such correlation can be measured, quantified, understood and used to make predictions on future touch position. Furthermore, this thesis also evaluates the predictive power of the sensor data. We show this through a number of studies. In Chapter 5 we show that probabilistic modelling of sensor inputs and recorded touch locations can be used to predict the general area of future touches on touchscreen. In Chapter 7, using SVM classifiers, we show that data from implicit sensing from general mobile interactions is user-specific. This can be used to identify users implicitly. In Chapter 6, we also show that touch interaction errors can be detected from sensor data. In our experiment, we show that there are sufficient distinguishable patterns between normal interaction signals and signals that are strongly correlated with interaction error. In all studies, we show that performance gain can be achieved by combining sensor inputs.

Efeito de 13 semanas de treinamento físico militar sobre a composição corporal e o desempenho físico dos alunos da escola preparatória de cadetes do exército

INTRODUCTION: Armies from all over the world acknowledge the importance of good physical fitness for the performance of military duties. The Military Exercise Training (MET) attempts to provide assistance to this search for better physical fitness and performance. OBJECTIVE: To verifying the effect on the body composition and physical performance of the students at ESPCEX (Military School for Preparation of Army Cadets) after 13 weeks of MET. METHODS: The sample was formed by 287 male students from the ESPCEX, whose average age was 18.33 ±1.26. Such students accomplished a boarding school routine, having defined schedules, meals and activities from which they were only released during the weekends. The MET was accomplished five days a week and it comprised both aerobic and resistance training. Measurement of body mass, height, skinfold (triceps, abdominal and suprailiac) was accomplished during pre and post training periods, and the following tests were performed: 12-minutes-run, oblique sit up, arm push up and pull up. Fat percentage, fat-free body mass and fat body mass were calculated using the anthropometric data based on the Guedes 3 skinfold protocol. RESULTS: Significant reduction in fat body mass, fat percentage and in triceptal and abdominal skinfold, as well as increase in suprailiac skinfold and fat-free body mass was observed when anthropometric and body composition data were compared, during the initial and the final periods of training. Significant improvement also occurred in all prformed physical tests, in which better performance was achieved. CONCLUSION: The acquired data suggest that performance of MET 5 days a week brought significantly improved body composition as well as physical performance

Perfil antropometrico e desempenho fisico-motor de crianças e jovens com deficiencia visual participantes do atletismo nos 1. Jogos Escolares da Confederação Brasileira de Desportos para Cegos

Universidade Estadual de Campinas . Faculdade de Educação Física

Análise transversal da estrutura óssea e parâmetros hematológicos em futebolistas profissionais = : Cross-sectional analysis of bone structure and hematological parameters in professional soccer players

Universidade Estadual de Campinas. Faculdade de Educação Física

Building a Open Source Framework for Virtual Medical Training

This paper presents a framework to build medical training applications by using virtual reality and a tool that helps the class instantiation of this framework. The main purpose is to make easier the building of virtual reality applications in the medical training area, considering systems to simulate biopsy exams and make available deformation, collision detection, and stereoscopy functionalities. The instantiation of the classes allows quick implementation of the tools for such a purpose, thus reducing errors and offering low cost due to the use of open source tools. Using the instantiation tool, the process of building applications is fast and easy. Therefore, computer programmers can obtain an initial application and adapt it to their needs. This tool allows the user to include, delete, and edit parameters in the functionalities chosen as well as storing these parameters for future use. In order to verify the efficiency of the framework, some case studies are presented.

INDUSCON 2010

The following papers constitute a selection among the best papers presented at the Ninth IEEE/IAS International Conference on Industry Applications (INDUSCON) held in Sao Paulo from 8(th) to 10(th) of November, 2010. This event gathered a significant number of people from academia and industry interested in applications of Electrical and Electronic engineering to industry.

Um site, um conteúdo e incontáveis dispositivos: as estratégias de design utilizadas para ambientes multiplataformas

Dissertação apresentada à Escola Superior de Comunicação Social como parte dos requisitos para obtenção de grau de mestre em Audiovisual e Multimédia.

Introducing numerical analysis tools in engineering: a Scilab user case in electronics course

Software tools in education became popular since the widespread of personal computers. Engineering courses lead the way in this development and these tools became almost a standard. Engineering graduates are familiar with numerical analysis tools but also with simulators (e.g. electronic circuits), computer assisted design tools and others, depending on the degree. One of the main problems with these tools is when and how to start use them so that they can be beneficial to students and not mere substitutes for potentially difficult calculations or design. In this paper a software tool to be used by first year students in electronics/electricity courses is presented. The growing acknowledgement and acceptance of open source software lead to the choice of an open source software tool – Scilab, which is a numerical analysis tool – to develop a toolbox. The toolbox was developed to be used as standalone or integrated in an e-learning platform. The e-learning platform used was Moodle. The first approach was to assess the mathematical skills necessary to solve all the problems related to electronics and electricity courses. Analysing the existing circuit simulators software tools, it is clear that even though they are very helpful by showing the end result they are not so effective in the process of the students studying and self learning since they show results but not intermediate steps which are crucial in problems that involve derivatives or integrals. Also, they are not very effective in obtaining graphical results that could be used to elaborate reports and for an overall better comprehension of the results. The developed tool was based on the numerical analysis software Scilab and is a toolbox that gives their users the opportunity to obtain the end results of a circuit analysis but also the expressions obtained when derivative and integrals calculations, plot signals, obtain vector diagrams, etc. The toolbox runs entirely in the Moodle web platform and provides the same results as the standalone application. The students can use the toolbox through the web platform (in computers where they don't have installation privileges) or in their personal computers by installing both the Scilab software and the toolbox. This approach was designed for first year students from all engineering degrees that have electronics/electricity courses in their curricula.

Aplicação da abordagem KAIZEN na produção de artigos de papelaria

No âmbito da unidade curricular Tese/Dissertação do 2ºano do Mestrado em Engenharia Eletrotécnica – Ramo Sistemas e Planeamento Industrial do Instituto Superior de Engenharia do Porto, o presente trabalho descreve o estágio curricular efetuado num projeto industrial de melhoria em parceria com o Kaizen Institute, uma empresa de consultoria operacional. Este projeto foi desenvolvido numa empresa de produção e redistribuição de artigos de papelaria e escritório, a Firmo AVS – Papeis e Papelaria,S.A.. O acordo efetuado entre o Kaizen Institute e a Firmo AVS foi o de promover e incutir a cultura da melhoria continua e da mudança de atitudes e comportamentos por parte dos colaboradores da Firmo, sendo que numa fase inicial o foco do projeto foi o departamento de produção de envelopes, designada por área piloto, expandindo-se posteriormente a metodologia Kaizen aos restantes departamentos. A realização deste projeto teve como objetivo a implementação de conceitos elementares de melhoria continua nomeadamente alguns pilares ou ferramentas do Total Flow Management (TFM) e do Kaizen Management System (KMS) na empresa Firmo, de forma a reduzir ou eliminar desperdícios, incremento do envolvimento dos colaboradores, melhoria da comunicação e trabalho em equipa, estandardização de processos produtivos, criação de normas de trabalho, utilização de ferramentas SMED para a redução de tempos improdutivos e aumento da produtividade. Várias foram as dificuldades presentes no terreno para a implementação destes objetivos mas com as diversas ferramentas e workshops realizados na organização, conseguiu-se o envolvimento de todos os colaboradores da organização e a obtenção de resultados satisfatórios nomeadamente ao nível da comunicação e trabalho em equipa, organização e limpeza dos postos de trabalho, standard work (normalização do trabalho), diminuição do lead time nos processos produtivos e consequente aumento de produtividade.

Transmisión de datos PDA - servidor web

En aquest projecte es visualitza la trajectòria d'un vehicle (aeri o terrestre) en una pàgina web. Per això es disposa d'una PDA (Personal Digital Assistant), en la qual es té informació actualitzada de la posició i de la velocitat d’aquest vehicle. Aquestes dades són obtingudes d'un sistema que combina la navegació inercial i el GPS (Global Position System), els quals estimen de manera precisa la trajectòria del vehicle. A fi d'oferir una visualització en temps real, versàtil, accessible i amigable a l'usuari de la trajectòria del vehicle, s'ha desenvolupat un sistema de visualització on-line que proporciona un millor rendiment en comparació amb la qual es venia fent en la PDA. Per a dur-lo a terme s'implementa una interfície d'usuari en la PDA que ens permet transmetre aquesta informació via WIFI a la pàgina web, d'igual forma al servidor web es crea una interfície que interpreta i gestiona aquestes dades per a posteriorment ser graficats a Google Maps.

Pangea, una xarxa al servei de l'educació

Pangea és una xarxa connectada a Internet. Per aixo, quan em van proposar d'escriure aquest article de seguida vaig pensar que en aquests moments és molt difícil fer referéncies a temes relacionats amb Internet sense caure en el parany deis topics, o sense contribuir a donar la imatge que actualment ja no es pot viure sense aquesta gran xarxa. Aquesta darrera percepció és la que em fa més por, sobretot en el camp de I'educació. La popularització d'lnternet en el món educatiu suposa i suposara en els propers anys una petita revolució que pot contribuir a la innovació en el sistema educatiu, pero malauradament els reptes de la renovació i millora de I'ensenyament en tots els seus nivells i aspectes no es poden abordar exclusivament davant d'un ordinador connectat a la xarxa.

EDUMAT: Fòrum sobre Educació Matemàtica. Una eina electrònica al servei de la comunitat de professionals de l'Educació Matemàtica.

EDUMAT va sorgir d'una iniciativa personal, en el marc del Departament de Didàctica de les Ciències Experimentals i de les Matemàtiques de la Universitat de Barcelona, el mes de novembre de 1998. No es tracta d'una iniciativa pionera. Dues llistes modèliques van precedir EDUMAT: l'una a la Universitat de Barcelona, que funciona des del 1995, el Fòrum sobre la Docència Universitària, coordinat per la professora Begoña Gros, que va ser presentat a Temps d'Educació, nº 16. 1, en l'àmbit de l'educació matemàtica, també des del 1995, funciona un altre fòrum internacional de debat coordinat per Peter Gates, de la Universitat de Nottingham, a Anglaterra. I les llistes de distribució a Internet es compten per milers.