abstract: The detection and segmentation of objects appearing in a natural scene, often referred to as Object Detection, has gained a lot of interest in the computer vision field. Although most existing object detectors aim to detect all the objects in a given scene, it is important to evaluate whether these methods are capable of detecting the salient objects in the scene when constraining the number of proposals that can be generated due to constraints on timing or computations during execution. Salient objects are objects that tend to be more fixated by human subjects. The detection of salient objects is important in applications such as image collection browsing, image display on small devices, and perceptual compression. This thesis proposes a novel evaluation framework that analyses the performance of popular existing object proposal generators in detecting the most salient objects. This work also shows that, by incorporating saliency constraints, the number of generated object proposals and thus the computational cost can be decreased significantly for a target true positive detection rate (TPR). As part of the proposed framework, salient ground-truth masks are generated from the given original ground-truth masks for a given dataset. Given an object detection dataset, this work constructs salient object location ground-truth data, referred to here as salient ground-truth data for short, that only denotes the locations of salient objects. This is obtained by first computing a saliency map for the input image and then using it to assign a saliency score to each object in the image. Objects whose saliency scores are sufficiently high are referred to as salient objects. The detection rates are analyzed for existing object proposal generators with respect to the original ground-truth masks and the generated salient ground-truth masks. As part of this work, a salient object detection database with salient ground-truth masks was constructed from the PASCAL VOC 2007 dataset. Not only does this dataset aid in analyzing the performance of existing object detectors for salient object detection, but it also helps in the development of new object detection methods and evaluating their performance in terms of successful detection of salient objects. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2019

Dans l'objectif d'améliorer et d'automatiser la chaîne de reproductiond'objet qui va de l'acquisition à l'impression 3D. Nous avons cherché à caractériserde la saillance sur les objets 3D modélisés par la structure d'un maillage 3D.Pour cela, nous avons fait un état de l'art des méthodes d'estimation des proprié-tés différentielles, à savoir la normale et la courbure, sur des surfaces discrètes sousla forme de maillage 3D. Pour comparer le comportement des différentes méthodes,nous avons repris un ensemble de critères de comparaison classique dans le domaine,qui sont : la précision, la convergence et la robustesse par rapport aux variations duvoisinage. Pour cela, nous avons établi un protocole de tests mettant en avant cesqualités. De cette première comparaison, il est ressorti que l'ensemble des méthodesexistantes présentent des défauts selon ces différents critères. Afin d'avoir une estimationdes propriétés différentielles plus fiable et précise nous avons élaboré deuxnouveaux estimateurs. / With the aim to improve and automate the object reproduction chainfrom acquisition to 3D printing .We sought to characterize the salience on 3D objectsmodeled by a 3D mesh structure. For this, we have a state of the art of estimatingdifferential properties methods, namely normal and curvature on discrete surfaces inthe form of 3D mesh. To compare the behavior of different methods, we took a set ofclassic benchmarks in the domain, which are : accuracy, convergence and robustnesswith respect to variations of the neighbourhood. For this, we have established atest protocol emphasizing these qualities. From this first comparision, it was foundthat all the existing methods have shortcomings as these criteria. In order to havean estimation of the differential properties more reliable and accurate we developedtwo new estimators.

The southern and central Appalachian foreland fold-thrust belt comprises a series of orogen -scale curves that extend from Alabama to New York. One of these is the Tennessee salient, a foreland-convex curve that extends from Cartersville, Georgia, to Roanoke, Virginia. Development of a kinematic model for deformation in the salient has been hindered by a paucity of penetrative deformation in this generally low temperature, low volume-loss portion of the orogen. Industry seismic reflection lines provide greater resolution of subsurface geometry of both the basement surface and the overlying fold-thrust belt, confirming some previous interpretations and changing others. A series of cross sections based on the seismic reflection data incorporates the improved understanding of basement geometry, as well as new interpretations of fold-thrust belt structures such as a sub-thrust detachment fold along the western margin of the Valley and Ridge, a smaller detachment fold along the Cumberland Escarpment, and a duplex below the Knoxville sheet in southeastern Tennessee. The cross sections, combined with recently published analyses of calcite twin strain and paleomagnetic data around the salient, provide sufficient data to develop a new palinspastic reconstruction method and to propose a kinematic model for development of the salient. The basis of the reconstruction method is, in areas where the front of the indenter is oriented oblique to transport, the maximum shortening direction and particle displacement paths are also oblique to the bulk transport direction. Cross sections, kinematic indicators, and palinspastic reconstructions suggest that the Tennessee salient is a primary arc formed by a combination of uniform displacement in a single direction and transport-parallel simple shear (plane strain), that most major faults formed initially curved in front of the irregularly shaped Blue Ridge-Inner Piedmont indenter, and that transport in the fold-thrust belt may have occurred by plan view movement on networks of minor faults, which permitted forelandward propagation of the curved faults without significant rotation. Although the technique does not provide a unique solution, the resulting palinspastic restoration is kinematically admissible and geometrically reasonable. So, it may improve palinspastic restorations of facies in basins with no vertical axis rotations and minimal penetrative strain. Attachments are in PDF format and may be opened with Adobe Reader™.

The researchers at the Department of Electricity at Uppsala University has recently entered the field of electric motor design, however no real knowledge of motor control of salient pole permanent magnet motors exists in the department. This thesis will present a general description of the control method of motors that exist today, this has been done by reviewing existing literature. The literature review has shown that there are at least three control methods with a significant different in their control approach, Scalar-, Field Oriented- and Direct Torque- Control. The two last methods were chosen by the author as the most useful and was implemented and simulated together with the newly developed motor in MATLAB Simulink to evaluate their performance. The simulation results show that there is no difference in performance of the two methods, but they show a difference in efficiency. The results show that it's worth to develop both methods further, mainly for reducing the torque and current ripple. This result was not expected according to literature, which suggests that the Field Oriented Control has a lower torque ripple. This could be caused by the choice of hysteresis control for inverter switching, instead of more sophisticated methods with a proportional integral derivative controller (PID) together with Sinusoidal Pulse Width Modulation (SPWM) or Space Vector Modulation (SVM).

Zircon (U-Th)/He (zircon He) dating has become a widely used thermochronologic method in the geosciences. Practitioners have traditionally interpreted (U-Th)/He dates from zircons across a broad spectrum of chemical compositions with a single set of ⁴He diffusion kinetics derived from only a handful of crystals (Reiners et al., 2004). However, it has become increasingly clear that a "one-size-fits-all" approach to these kinetics is inadequate, leading to erroneous conclusions and incongruent data. This dissertation develops a more grain-specific approach by showing the fundamental role that intracrystalline radiation damage plays in determining the He diffusivity in a given zircon. I present three appendices that seek to quantify the radiation damage effect on He diffusion in zircon, explain how this effect manifests in zircon He dates, and show how to exploit such manifestations to better constrain sample thermal histories. Of particular importance, this dissertation represents the first comprehensive study to concentrate on the entire damage spectrum found in natural zircon and also the first to show that two different mechanisms affect He diffusion in zircon in different ways across this spectrum. In the first appendix, I and my fellow co-authors describe results from a series of step-heating experiments that show how the alpha dose of a given zircon, which we interpret to be correlated with accumulated radiation damage, influences its He diffusivity. From 1.2 × 10¹⁶ α/g to 1.4 × 10¹⁸ α/g, He diffusivity at a given temperature decreases by three orders of magnitude, but as alpha dose increases from ~2 × 10¹⁸ α/g to 8.2 × 10¹⁸ α/g, He diffusivity then increases by about nine orders of magnitude. We parameterize both the initial decrease and eventual increase in diffusivity with alpha dose with a function that describes these changes in terms of increasing abundance and size of intracrystalline radiation damage zones and resulting effects on the tortuosity of He migration pathways and dual-domain behavior. This is combined with another equation that describes damage annealing in zircon. The end result is a new model that constrains the coevolution of damage, He diffusivity, and He date in zircon as a function of its actinide content and thermal history. The second and third appendices use this new model to decipher zircon He datasets comprising many single grain dates that are correlated with effective uranium (eU, a proxy for the relative degree of radiation damage among grains from the same sample). The model is critical for proper interpretation of results from igneous settings that show date-eU correlations and were once considered spurious (appendix B). When applied to partially reset sedimentary rocks, other sources of date variability, such as damage and He inheritance, have to be considered as well (appendix C).

The requirements on vehicle industry are constantly getting stricter, especially when it comes to emissions. At the same time cars, trucks and buses are needed for our way of living. This have forced companies to be as ecient as possible in their way of using fossil fuels for travelling and transport. To increase the eciency companies investigate all possible fuel savings to decrease their carbon footprint as much as possible. One area of savings that is not that obvious to many people is the alternator. Several percent of the total energy used by a vehicle are needed to operate the alternator. With a typical alternator eciency of 70% considerable savings can be achieved. This thesis that concern alternator eciency was carried out at Scania in Södertälje, Sweden. The goal of the thesis is to construct a mathematical model of an alternator. The model is supposed to consider all losses in the alternator and together with the output power give an eciency model of the alternator at different speeds and loads. A great part of the project has been dealing with the magnetic losses. The magnetic losses have been modeled as an equivalent circuit with the load angle as a central piece. The equivalent circuit is built up by the fact that the alternator used in the vehicles is a salient pole alternator. The equivalent circuit describes a voltage equation where the voltage drop over the magnetic inductance is described. From that relations between the signals in the alternator and output signals can be written. The alternator model is then used together with data recorded from different buses all over the world, this to be able to investigate how the alternator contributes to the fuel consump- tion depending on the way that the buses are driven. The result of this thesis is a mathematical model that describes the losses in the alternator for different load cases and speeds.

Σκοπός της παρούσας διπλωματικής εργασίας είναι η ανάλυση της λειτουργίας τριφασικής σύγχρονης γεννήτριας εκτύπων πόλων κατά την εκκίνηση, καθώς και την ηλεκτρομαγνητική μελέτη ενός συστήματος δύο βαθμών ελευθερίας (στρόβιλος-γεννήτρια). Τέλος, θα γίνει σχολιασμός των αποτελεσμάτων. / The aim of the thesis is to study and analyze the operation of three phase synchronous generators with salient poles at startup, and electromechanical advancement of a system of two degrees of freedom (turbinegenerator) which is oscillated. Comments of the findings will follow in the last chapter. The thesis is structured in five chapters. Chapter 1 introduces the literature review where there is a general description of the basics of electric energy and the principles regarding the construction and operation of synchronous machines and in particular three-phase synchronous machines with salient poles. In chapter 2, a comprehensive analysis of oscillating systems is made of one and two degrees of freedom. A comparison is then made between linear and rotary motion. In the end the mathematical model of a rotating part with two degrees of freedom and the constant of twisted spring is presented. Chapter 3 shows the main steps followed during the design stage of the model and it includes images of the main parts of the generator and the control windows of the software Opera-2d. In Chapter 4 we make a reference to the PI-controller which was used to control the speed of the generator. Then we give the parameters of analysis. In Chapter 5 shows the presentation of the simulation results during the startup of the synchronous generator until the steady state.

Visual saliency computation is about detecting and understanding salient regions and elements in a visual scene. Algorithms for visual saliency computation can give clues to where people will look in images, what objects are visually prominent in a scene, etc. Such algorithms could be useful in a wide range of applications in computer vision and graphics. In this thesis, we study the following visual saliency computation problems. 1) Eye Fixation Prediction. Eye fixation prediction aims to predict where people look in a visual scene. For this problem, we propose a Boolean Map Saliency (BMS) model which leverages the global surroundedness cue using a Boolean map representation. We draw a theoretic connection between BMS and the Minimum Barrier Distance (MBD) transform to provide insight into our algorithm. Experiment results show that BMS compares favorably with state-of-the-art methods on seven benchmark datasets. 2) Salient Region Detection. Salient region detection entails computing a saliency map that highlights the regions of dominant objects in a scene. We propose a salient region detection method based on the Minimum Barrier Distance (MBD) transform. We present a fast approximate MBD transform algorithm with an error bound analysis. Powered by this fast MBD transform algorithm, our method can run at about 80 FPS and achieve state-of-the-art performance on four benchmark datasets. 3) Salient Object Detection. Salient object detection targets at localizing each salient object instance in an image. We propose a method using a Convolutional Neural Network (CNN) model for proposal generation and a novel subset optimization formulation for bounding box filtering. In experiments, our subset optimization formulation consistently outperforms heuristic bounding box filtering baselines, such as Non-maximum Suppression, and our method substantially outperforms previous methods on three challenging datasets. 4) Salient Object Subitizing. We propose a new visual saliency computation task, called Salient Object Subitizing, which is to predict the existence and the number of salient objects in an image using holistic cues. To this end, we present an image dataset of about 14K everyday images which are annotated using an online crowdsourcing marketplace. We show that an end-to-end trained CNN subitizing model can achieve promising performance without requiring any localization process. A method is proposed to further improve the training of the CNN subitizing model by leveraging synthetic images. 5) Top-down Saliency Detection. Unlike the aforementioned tasks, top-down saliency detection entails generating task-specific saliency maps. We propose a weakly supervised top-down saliency detection approach by modeling the top-down attention of a CNN image classifier. We propose Excitation Backprop and the concept of contrastive attention to generate highly discriminative top-down saliency maps. Our top-down saliency detection method achieves superior performance in weakly supervised localization tasks on challenging datasets. The usefulness of our method is further validated in the text-to-region association task, where our method provides state-of-the-art performance using only weakly labeled web images for training.

While pragmatics, through its integration of situational awareness and real world relevant knowledge, offers a high level of analysis that is suitable for real interpretation of natural dialogue, semantics, on the other end, represents a lower yet more tractable and affordable linguistic level of analysis using current technologies. Generally, the understanding of semantic meaning in literature has revolved around the famous quote ``You shall know a word by the company it keeps''. In this thesis we investigate the role of context constituents in decoding the semantic meaning of the engulfing context; specifically we probe the role of salient concepts, defined as content-bearing expressions which afford encyclopedic definitions, as a suitable source of semantic clues to an unambiguous interpretation of context. Furthermore, we integrate this world knowledge in building a new and robust unsupervised semantic model and apply it to entail semantic relatedness between textual pairs, whether they are words, sentences or paragraphs. Moreover, we explore the abstraction of semantics across languages and utilize our findings into building a novel multi-lingual semantic relatedness model exploiting information acquired from various languages. We demonstrate the effectiveness and the superiority of our mono-lingual and multi-lingual models through a comprehensive set of evaluations on specialized synthetic datasets for semantic relatedness as well as real world applications such as paraphrase detection and short answer grading. Our work represents a novel approach to integrate world-knowledge into current semantic models and a means to cross the language boundary for a better and more robust semantic relatedness representation, thus opening the door for an improved abstraction of meaning that carries the potential of ultimately imparting understanding of natural language to machines.

A holistic evolutionary approach is used in the Gorongosa thesis in which emphasis is on the salient reciprocal relations and kinetic succession of land surfaces and biotic communities, influenced by landscape processes and prime mover components. As correlations of these relations and processes require both a total interacting framework and the details of its prime components, the thesis is divided into three main parts: (1) synopsis of the essence of the Gorongosa ecosystem and the approach used in field ecology (Perspective); (2) correlation of the physical and living components of the ecosystem; and (3) synthesis. The study attempts to relate the salient features of processes and correlations into a co-evolutionary whole, caught at that particular stage in space and time by the study. The chapter titled Process and Response is the central pivot of the thesis combining the kinetic aspects of geomorphological landscape changes with co-evolutionary sequences of biotic communities which change (expand, contract, and recombine) kaleidoscopically in space and time, in appearance and content. The prime movers in ecosystem change are on the physical side, nickpoint headward eroding sequences and edaphic change in soil moisture balance, and on the biotic side, the frugivores and large ungulate components which affect geomorphic and habitat modification are central. Of these, soil moisture appears to be the master factor. All climatic influences too, appear to be expressed through the edaphic controls which change in-situ, or with each geomorphic surface replacement sequence. This factor seems to orchestrate the opportunities and constraints from below on the possible community evolution possible in a particular time and place. From this, a template of salient factors of the Gorongosa ecosystem is provided for management, based on causes and trends in the kinetic evolution of the various ecosystems. To maintain a diversity of ecosystems in Gorongosa, the fundamental management action is to reinforce or reinstate the natural local base level sills which cause ponding of floodwaters responsible for the mosaic of grasslands and slack marshes of high primary productivity and ungulate carrying capacity. Concomittantly reductions of certain overpopulated ungulate species, chiefly hippo, are required so that management is effective. As natural processes are dynamic, it is necessary to identify and evaluate those salient factors operating at a particular time, as these key controls are altered and replaced by others through natural kinetic succession of landscapes and biotic communities. The salient factors governing the dynamics of an ecosystem or community thus require to be mapped at intervals, to provide templates of the trends and changing importance of key and master factors, in order to anticipate or predict what will result from their influences. With these data valid evaluation can be made with the other correlated information for meaningful management action. / Thesis (DSc)--University of Pretoria, 1977. / Zoology and Entomology / unrestricted