This thesis presents a framework for automatic recognition of human actions in un- controlled, realistic video data with fixed cameras, such as surveillance videos. In this thesis, we divide human action recognition into three steps: description, representation, and classification of local spatio-temporal features. The bag-of-features model was used to build the classifier. Fisher Vectors were also studied. We focus on the potential of the methods, with the joint optimization of two constraints: the classification precision and its efficiency. On the performance side, a new local descriptor, called Gradient Boundary Histograms (GBH), is adopted. It is built on simple spatio-temporal gradients, which can be computed quickly. We demonstrate that GBH can better represent local structure and motion than other gradient-based descriptors, and significantly outperforms them on large datasets. Our evaluation shows that compared to HOG descriptors, which are based solely on spatial gradient, GBH descriptor preserves the recognition precision even in difficult situation. Since surveillance video captured with fixed cameras is the emphasis of our study, removing the background before action recognition is helpful for improving efficiency. We first preprocess the video data by applying HOG to detect humans. GBH descriptor is then used at reduced spatial resolutions, which yields both high efficiency and low memory usage; in addition, we apply PCA to reduce the feature dimensions, which results in fast matching and an accelerated classification process. Experiments our methods achieved good performance in recognizing precision, while simultaneously highlighting effectiveness and efficiency.

It is well-established that the presentation of a startling acoustic stimulus (SAS) simultaneous with the go-signal in a simple reaction time (RT) task results in significant RT reductions, while leaving movement kinematics essentially unaltered. While this phenomenon, termed the StartReact effect, has been extensively studied, cortical involvement in the neural mechanism underlying the RT-facilitation effects of a SAS remains widely debated. Applying sub-threshold TMS to motor areas results in increased cortical excitability and reductions in control RT. When this technique was used in a startle paradigm no RT benefits were seen, providing evidence that the cortex may not be involved in the StartReact effect; however, these results may also have been due to a floor effect of startle RT. It has been shown that RT in response to a SAS is significantly slower for complex movements, providing a possible method of distinguishing between these hypotheses. As such, the purpose of the experiments in this thesis was to determine if the application of sub-threshold TMS following a SAS when preparing to react with a complex movement would facilitate startle RT. If so, it would provide evidence for cortical involvement in the RT-facilitation effects of startle. The first experiment revealed that the task employed did not lead to an increase in RT in startle conditions, limiting the ability to make conclusions regarding the StartReact effect. In the second experiment the timing complexity of the task was increased, with the goal of increasing startle RT; however, startle RT was again not significantly slower for the complex movement than the simple movement. Furthermore, there was again no effect of TMS stimulation condition on startle RT. These results suggest that either the cortex does not play a role in the StartReact effect, or a floor effect of RT was reached in startle conditions; thus, alternative methods of investigating the neural mechanism underlying the RT-facilitation effects of startle are warranted.

Pour le présent mémoire de maîtrise, nous avons retenu le roman de J.M.G. Le Clézio, Désert, dans lequel nous avons essayé d'analyser le temps et l'espace. Plus brièvement nous avons examiné la fonction des personnages dans ce même roman. Au cours de notre recherche, nous avons appris que l'écriture est le point de départ de la pensée de Le Clézio. La première dimension temporelle du roman est celle de l'histoire. Désert est un roman binaire. Il est divisé en deux grandes parties : l'histoire des hommes bleus et l'histoire de Lalla, jeune immigrée de dix-sept ans. La première est racontée au passé, la deuxième au présent. Le roman n'est pas construit chronologiquement, nous avons choisi le terme de mosaïque pour définir le découpage passë-présent-passé. La composition et le mode narratif permettent d'exprimer le temps ; deux niveaux temporels : le temps de la fiction et le temps de la narration. La multiplicité temporelle, c'est-à-dire temps de l'histoire et temps du récit a constitué une partie de notre étude. Le temps nous a conduit à l'espace qui s'est avéré un point essentiel de notre étude. Il n'est pas seulement le lieu où se joue l'histoire. Espace et description sont indissociables. Nous remarquons que la description chez Le Clézio est étroitement liée à la conscience et à l'état d'âme des personnages. L'espace contribue à l'élaboration de la narration. Sous des formes diverses, la description suggère la découverte de certains mythes dont nous avons analysé la valeur symbolique à travers les images présentées. La fonction des personnages n'est pas dissociable du reste de notre analyse. Lieux et descriptions expliquent les personnages. A travers leur caractère nous retrouvons uneanalyse psychologique et sociale. / Arts, Faculty of / French, Hispanic, and Italian Studies, Department of / Graduate

There has been little published about the characteristics of computer jobs running on modern time-sharing computer systems, due largely to the lack of appropriate programs and equipment necessary to measure the parameters involved. In this thesis, measures are presented for some of the important characteristics of jobs. The Data Collection Facility, which is part of the Michigan Terminal System, was used to this end. The Michigan Terminal System is a time-sharing operating system for the IBM 360/67 computer, and supports batch and terminal users simultaneously. Chapter 1 gives an outline of the problem, and other work which has been done in this line. It also contains a reasonably detailed description of the Michigan Terminal System. In Chapter 2, measurements of requested CPU service, CPU service obtained, system and user response times, I/O delays, and page waiting times are given. Chapter 3 outlines the storage requirements of jobs, and gives a model which will generate profiles of storage required by jobs over their running times, which are very similar to profiles observed for actual jobs. Some discussion of the results is given in chapter 4, and also a simple model of the system is shown which might be used in a simulation study employing measurements taken in this study. / Science, Faculty of / Computer Science, Department of / Graduate

In this work, methods for on-line identification of discrete-time systems and for parameter tracking of sampled-data systems are presented. These methods are suitable for implementation using small computers. A class of methods for the identification of the coefficients of linear and nonlinear difference equations is developed. The philosophy of identification is divided into three parts based on the norm of the error to be minimized. Techniques are derived using a common framework of minimization of these error functions, incorporating uniqueness and stability properties. Practical examples are included which demonstrate that among these proposed methods the identification error method solves the problem successfully. Extensions of these methods to continuous systems are briefly outlined. Methods are proposed for the generation of parameter sensitivity functions for sampled-data systems on a hybrid or a digital computer. Both linear and nonlinear systems are considered, and for a class of linear, systems,, an economical approach for the generation is developed, making extensive use of signal flow graph techniques. A new technique is devised for solving the problem of parameter tracking of linear and nonlinear sampled-data systems using the sensitivity functions. Examples are presented to demonstrate that the proposed techniques solve the problem. / Applied Science, Faculty of / Electrical and Computer Engineering, Department of / Graduate

In this study a set of rules that generate certain time expressions in English is constructed. The methodology used is mainly that outlined by Noam Chomsky in Aspects of the Theory of Syntax (1965). The discussion is confined to those time expressions that are single words, single phrases, or sequences of phrases in surface structure. These have the basic deep structure Prep + Det + N (S'), where N has the syntactic feature [+ Time]. Surface structure single word time expressions are derived from this deep structure by deleting Prep and rewriting the NP as a single lexical item; surface structure sequences of phrases are derived by applying the relative clause transformation to the embedded S. Chomsky's list of syntactic features for nouns is extended by adding some inherent features and some selectional features. The additions are needed to distinguish nouns that can occur in time expressions from those that cannot, to state the collocation restrictions between some prepositions and determiners and the time nouns, and to state certain ordering restrictions on surface structure sequences of phrases. The time expressions considered are subcategorized into Locative Time and Duration Time on the basis of collocation with some subclasses of Verb. These subcategories are formally distinct in that their prepositions are mutually exclusive. Locative Time expressions are further sub-categorized into Dynamic Time expressions, those whose collocation restrictions with Auxiliary expansions are linguistically determinable, and Static Time expressions, those whose collocation restrictions are not linguistically determinable. These two subcategories are, again, formally distinct, for their determiners are mutually exclusive. / Arts, Faculty of / English, Department of / Graduate

The building for the school of the Built Environment focuses on the concept of time as form giver in architecture. Time, a governing factor of existence, regulating simultaneously through a linear and cyclical pattern in its operation. The aim of the discourse is to establish a parity between humankind and nature within an urban environment. The goal is to create an environment that acts as time-mediator between the metaphysical and physical city and its myriad users. The emphasis being on the user-interface on the project and its surroundings. Thereby potentially establishing a platform where the city is continually challenged in terms of observing and being observed. The architecture itself should provide an abridgement of moments in time. The prominence of the proposed site opens up the possibility of investigating an iconic branding image for the campus. This could be achieved in terms of potential visual resource and movement pattern. Copyright 2005, University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. Please cite as follows: Boer, GP 2005, A study in Time, MArch(Prof) dissertation, University of Pretoria, Pretoria, viewed yymmdd < http://upetd.up.ac.za/thesis/available/etd-02032005-080823 / > / Dissertation (MArch (Prof))--University of Pretoria, 2006. / Architecture / unrestricted

Transit agencies have the opportunity to improve the delivery of services by using data from Intelligent Transportation Systems (ITS). On-time performance is an important measure. The objective of this paper is to adjust the timetables so that the probability of on-time performance is maximized. For this purpose we analyze data distributions of travel time and also consider the general case that data distribution is unknown. Statistical procedures are presented to find scheduled time for some selected distributions. Monte Carlo simulation is introduced for the purpose of finding scheduled time when data distribution is not known. Simulation studies indicate that the on-time performance would increase using the proposed methodology. The contribution of this paper is to provide transit system a procedure to set up or update their timetables based on current ITS data and its distribution, and hence increase level of service.

Orientador: Manuel Folledo / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Matematica, Estatistica e Ciencia da Computação / Made available in DSpace on 2018-07-20T07:03:36Z (GMT). No. of bitstreams: 1 Alves_JoaoMurta_M.pdf: 4251948 bytes, checksum: 7a7ce34fca72e933635ea44d7f83a346 (MD5) Previous issue date: 1995 / Resumo: Trata-se de um estudo da filosofia Just In Time (JIT) de administração da manufatura, focalizando o aprimoramento do processo produtivo - em ganhos de qualidade e produtividade - como estratégia para ajudar uma empresa a alcançar vantagem competitiva em custo. A abordagem é sistêmica, partindo da ênfase da necessidade de se elaborar uma estratégia bem fundamentada no conhecimento da indústria e do seu meio ambiente, passando pelo conhecimento do sistema Just In Time e seus principais objetivos, conceituando o custo real (valor agregado) e definindo os indicadores de produtividade e qualidade. Em seguida identificam-se os desperdícios da produção e apresentam-se as ferramentas do Just In Time para combate-los, dando uma atenção especial ao sistema kanban e ao sistema da qualidade. O estudo de caso na Ericsson Telecomunicações contribui para melhor compreender a implementação e eficácia do sistema JIT. / Abstract : This work studies the Just in Time (JIT) philosophy of manufacturing administration, focusing the productive process improvement - in terms of productivity and quality gains - as a strategy to support a company to reach competitive advantage in cost. The adopted approach is systemic, initiating by emphasizing the necessity of elaborating a strategy very well based, on the knowledge of the industry and its operational environment, continuing with the application of the Just fn Time principles and its main objectives, conceptualizing the real cost (aggregated value) and defining the productivity and quality indicators. Finally, the study identifies the production wastefulness and the Just in Time tools to defeat them, with special attention on the Kanban system and on the quality system. The Ericsson Telecommunications case study contributes for a better understanding of the JIT system implementation and efficacy. / Mestrado / Mestre em Qualidade

No description available.