Nous présentons dans cette thèse un système de correction automatique d'annotations (tags) fournies par des utilisateurs qui téléversent des vidéos sur des sites de partage de documents multimédia sur Internet. La plupart des systèmes d'annotation automatique existants se servent principalement de l'information textuelle fournie en plus de la vidéo par les utilisateurs et apprennent un grand nombre de "classifieurs" pour étiqueter une nouvelle vidéo. Cependant, les annotations fournies par les utilisateurs sont souvent incomplètes et incorrectes. En effet, un utilisateur peut vouloir augmenter artificiellement le nombre de "vues" d'une vidéo en rajoutant des tags non pertinents. Dans cette thèse, nous limitons l'utilisation de cette information textuelle contestable et nous n'apprenons pas de modèle pour propager des annotations entre vidéos. Nous proposons de comparer directement le contenu visuel des vidéos par différents ensembles d'attributs comme les sacs de mots visuels basés sur des descripteurs SIFT ou des motifs fréquents construits à partir de ces sacs. Nous proposons ensuite une stratégie originale de correction des annotations basées sur la fréquence des annotations des vidéos visuellement proches de la vidéo que nous cherchons à corriger. Nous avons également proposé des stratégies d'évaluation et des jeux de données pour évaluer notre approche. Nos expériences montrent que notre système peut effectivement améliorer la qualité des annotations fournies et que les motifs fréquents construits à partir des sacs de motifs fréquents sont des attributs visuels pertinents / This thesis presents a new system for video auto tagging which aims at correcting the tags provided by users for videos uploaded on the Internet. Most existing auto-tagging systems rely mainly on the textual information and learn a great number of classifiers (on per possible tag) to tag new videos. However, the existing user-provided video annotations are often incorrect and incomplete. Indeed, users uploading videos might often want to rapidly increase their video’s number-of-view by tagging them with popular tags which are irrelevant to the video. They can also forget an obvious tag which might greatly help an indexing process. In this thesis, we limit the use this questionable textual information and do not build a supervised model to perform the tag propagation. We propose to compare directly the visual content of the videos described by different sets of features such as SIFT-based Bag-Of-visual-Words or frequent patterns built from them. We then propose an original tag correction strategy based on the frequency of the tags in the visual neighborhood of the videos. We have also introduced a number of strategies and datasets to evaluate our system. The experiments show that our method can effectively improve the existing tags and that frequent patterns build from Bag-Of-visual-Words are useful to construct accurate visual features

Un nombre croissant d’applications graphiques 3D ont un impact sur notre société. Ces applications sont utilisées dans plusieurs domaines allant des produits de divertissement numérique, la conception assistée par ordinateur, aux applications médicales. Dans ce contexte, un moteur de recherche d’objets 3D avec de bonnes performances en résultats et en temps d’exécution devient indispensable. Nous proposons une nouvelle méthode pour l’indexation de modèles 3D basée sur des courbes fermées. Nous proposons ensuite une amélioration de notre méthode pour l’indexation partielle de modèles 3D. Notre approche commence par la définition d’une nouvelle fonction d’application invariante. Notre fonction d’application possède des propriétés importantes : elle est invariante aux transformations rigides et non rigides, elle est insensible au bruit, elle est robuste à de petits changements topologiques et elle ne dépend pas de paramètres. Cependant, dans la littérature, une telle fonction qui respecte toutes ces propriétés n’existe pas. Pour respecter ces propriétés, nous définissons notre fonction basée sur la distance de diffusion et la distance de migration pendulaire. Pour prouver les propriétés de notre fonction, nous calculons le graphe de Reeb de modèles 3D. Pour décrire un modèle 3D complet, en utilisant notre fonction d’application, nous définissons des courbes de niveaux fermées à partir d’un point source détecté automatiquement au centre du modèle 3D. Chaque courbe décrit alors une région du modèle 3D. Ces courbes créent un descripteur invariant à différentes transformations. Pour montrer la robustesse de notre méthode sur différentes classes de modèles 3D dans différentes poses, nous utilisons des objets provenant de SHREC 2012. Nous comparons également notre approche aux méthodes de l’état de l’art à l’aide de la base SHREC 2010. Pour l’indexation partielle de modèles 3D, nous améliorons notre approche en utilisant la technique sacs de mots, construits à partir des courbes fermées extraites, et montrons leurs bonnes performances à l’aide de la base précédente / A growing number of 3D graphic applications have an impact on today’s society. These applications are being used in several domains ranging from digital entertainment, computer aided design, to medical applications. In this context, a 3D object search engine with a good performance in time consuming and results becomes mandatory. We propose a novel approach for 3D-model retrieval based on closed curves. Then we enhance our method to handle partial 3D-model retrieval. Our method starts by the definition of an invariant mapping function. The important properties of a mapping function are its invariance to rigid and non rigid transformations, the correct description of the 3D-model, its insensitivity to noise, its robustness to topology changes, and its independance on parameters. However, current state-of-the-art methods do not respect all these properties. To respect these properties, we define our mapping function based on the diffusion and the commute-time distances. To prove the properties of this function, we compute the Reeb graph of the 3D-models. To describe the whole 3D-model, using our mapping function, we generate indexed closed curves from a source point detected automatically at the center of a 3D-model. Each curve describes a small region of the 3D-model. These curves lead to create an invariant descriptor to different transformations. To show the robustness of our method on various classes of 3D-models with different poses, we use shapes from SHREC 2012. We also compare our approach to existing methods in the state-of-the-art with a dataset from SHREC 2010. For partial 3D-model retrieval, we enhance the proposed method using the Bag-Of-Features built with all the extracted closed curves, and show the accurate performances using the same dataset

Essenslieferungen, Fahrdienste oder etwa die Erledigung von Kleinstaufträgen sog. „Microjobs“, besonders seit der COVID-19-Pandemie erfolgt diese Arbeit immer häufiger unter Vermittlung digitaler Plattformen. Diese Untersuchung eröffnet den Blick auf eine in der Vergangenheit gänzlich unbekannte Form der Arbeitsorganisation, bei der einer vermeintlichen Autonomie der Plattformarbeiter ein Intermediär gegenübersteht, der seine zentrale Position zur Steuerung und Kontrolle ebendieser Plattformarbeiter nutzt und dennoch das Vorliegen eines Arbeitsverhältnisses in der Regel vehement bestreitet. Die Dissertation arbeitet zunächst die theoretischen Grundlagen und wirtschaftlichen Hintergründe derartiger Plattformarbeit heraus. Im zweiten Schritt erfolgt auf Basis von Fallbeispielen – Clickworker, Deliveroo, Upwork – erstmals eine umfassende steuerrechtliche Einordnung des Phänomens Plattformarbeit. Hierbei werden die wesentlichen Besonderheiten im Vergleich zu tradierten Arbeitsverhältnissen, insbesondere die Steuerung der Plattformarbeiter mittels algorithmusbasierter Methoden der Verhaltensökonomie, und deren Auswirkungen auf die steuerliche Einordnung ausführlich beleuchtet. Es kann nachgewiesen werden, dass abhängig von der Art der zu erledigenden Aufgaben vom Intermediär ein Anreizsystem geschaffen werden muss, welches eine indirekte Steuerung des Plattformarbeiters zum Ziel und den Bezug von Einkünften aus nichtselbständiger Arbeit zur Folge hat. Abschließend werden für die ermittelten Unzulänglichkeiten der tradierten steuerlichen Abgrenzungskriterien Lösungsvorschläge entwickelt, etwa eine Beweislastregelung zu Gunsten der Plattformarbeiter. Auf Grund des Querschnittcharakters des Themas schafft die Arbeit zugleich interessante Ansatzpunkte für andere Rechtsgebiete, beispielsweise das Arbeits- oder Vertragsrecht. Stand des Werkes ist Juli 2020.

This master thesis deals with usage of machine learning methods in computer vision for classification of unknown images. The first part contains research of available machine learning methods, their limitations and also their suitability for this task. The second part describes the processes of creating training and testing gallery. In the practical part, the solution for the problem is proposed and later realised and implemented. Proper testing and evaluation of resulting system is conducted.

Skatten på engångsplastbärkassar som implementerades i Sverige år 2020 möttes med ideologiskt motstånd. Den forskning som gör gällande skatt på plastbärkassar finns främst i internationell kontext och är fokuserad på konsumenters beteenden och reaktioner. Denna studie gör gällande hur berörda verksamheter och organisationer resonerar kring skattens införande, samt dess resultat. Studien syftar också till att identifiera hinder och drivkrafter för implementeringen. För att undersöka förhållandet användes en explorativ ansats där intervjuer med en variation av berörda verksamheter utgjorde materialet för studien. Resultaten visar på att implementeringen mötts av missnöje av hälften av deltagarna på grund av skattens singulära syfte, samtidigt som den andra hälften anser att tillämpningsområdet var tillfredsställande. De hinder som identifierades var svårigheter att definiera engångs-, respektive flergångskassar, samt bristen på synkronisering eller kombination med andra styrmedel. De drivkrafter som identifierades relaterade till organiseringen och kommunikationen mellan de berörda verksamheterna, till trots förbättringsmöjligheter för dessa aspekter. En av slutsatserna är att styrmedel som detta bör nyttja både ett teknocentriskt perspektiv om plastbärkassens miljö-, och klimatpåverkan, samt ett socioekonomiskt perspektiv utgående från berörda verksamheter och organisationers behov. / The Swedish tax on plastic carrier bags that was implemented in 2020 was initially met with dissatisfaction from the public and stakeholders. Current research investigating the tax on plastic carrier bags was done in other countries and is focused on the reactions and behaviours of consumers. Thus, this study investigates stakeholders’ reasoning around the implementation of the tax as well as its results. Another aim is to identify obstacles and driving forces for the implementation. To do so, an explorative approach in combination with interviews of stakeholders were used. The results show that implementation was met with dissatisfaction by half of the participants in the study, owing to the singular aim of the tax. The other half of participants viewed the tax purpose as satisfactory. The obstacles that were identified was difficulties in distinguishing single-use from multi-use plastic carrier bags, as well as the lack of synchronization or combination with other measures. The driving forces that were identified related to the organization and communication between stakeholder, despite opportunities for improvement. One of the conclusions of the study was that policy measures like this tax should use a technocentric perspective on the environmental impact of the plastic carrier bag, in combination with a socioeconomic perspective on the needs of stakeholders.

This thesis compares hand-designed features with features learned by feature learning methods in video classification. The features learned by Principal Component Analysis whitening, Independent subspace analysis and Sparse Autoencoders were tested in a standard Bag of Visual Word classification paradigm replacing hand-designed features (e.g. SIFT, HOG, HOF). The classification performance was measured on Human Motion DataBase and YouTube Action Data Set. Learned features showed better performance than the hand-desined features. The combination of hand-designed features and learned features by Multiple Kernel Learning method showed even better performance, including cases when hand-designed features and learned features achieved not so good performance separately.

With the release of the Kinect device new possibilities appeared, allowing a simple use of image depth in image processing. The aim of this thesis is to propose a method for object detection and recognition in a depth map. Well known method Bag of Words and a descriptor based on Spin Image method are used for the object recognition. The Spin Image method is one of several existing approaches to depth map which are described in this thesis. Detection of object in picture is ensured by the sliding window technique. That is improved and speeded up by utilization of the depth information.

This paper addresses the problem of mobile robot localization based on current 2D and 3D data and previous records. Focusing on practical loop detection in the trajectory of a robot. The objective of this work was to evaluate current methods of image processing and depth data for issues of localization in environment. This work uses Bag of Words for 2D data and environment of point cloud with Viewpoint Feature Histogram for 3D data. Designed system was implemented and evaluated.

MSc (Microbiology) / See the attached abstract below

<div><b>1. A Closed-Loop Machine Learning and Compensation Framework for Geometric Accuracy Control of 3D Printed Products</b></div><div><b><br></b></div>Additive manufacturing (AM) systems enable direct printing of three-dimensional (3D) physical products from computer-aided design (CAD) models. Despite the many advantages that AM systems have over traditional manufacturing, one of their significant limitations that impedes their wide adoption is geometric inaccuracies, or shape deviations between the printed product and the nominal CAD model. Machine learning for shape deviations can enable geometric accuracy control of 3D printed products via the generation of compensation plans, which are modifications of CAD models informed by the machine learning algorithm that reduce deviations in expectation. However, existing machine learning and compensation frameworks cannot accommodate deviations of fully 3D shapes with different geometries. The feasibility of existing frameworks for geometric accuracy control is further limited by resource constraints in AM systems that prevent the printing of multiple copies of new shapes.<div><br></div><div>We present a closed-loop machine learning and compensation framework that can improve geometric accuracy control of 3D shapes in AM systems. Our framework is based on a Bayesian extreme learning machine (BELM) architecture that leverages data and deviation models from previously printed products to transfer deviation models, and more accurately capture deviation patterns, for new 3D products. The closed-loop nature of compensation under our framework, in which past compensated products that do not adequately meet dimensional specifications are fed into the BELMs to re-learn the deviation model, enables the identification of effective compensation plans and satisfies resource constraints by printing only one new shape at a time. The power and cost-effectiveness of our framework are demonstrated with two validation experiments that involve different geometries for a Markforged Metal X AM machine printing 17-4 PH stainless steel products. As demonstrated in our case studies, our framework can reduce shape inaccuracies by 30% to 60% (depending on a shape's geometric complexity) in at most two iterations, with three training shapes and one or two test shapes for a specific geometry involved across the iterations. We also perform an additional validation experiment using a third geometry to establish the capabilities of our framework for prospective shape deviation prediction of 3D shapes that have never been printed before. This third experiment indicates that choosing one suitable class of past products for prospective prediction and model transfer, instead of including all past printed products with different geometries, could be sufficient for obtaining deviation models with good predictive performance. Ultimately, our closed-loop machine learning and compensation framework provides an important step towards accurate and cost-efficient deviation modeling and compensation for fully 3D printed products using a minimal number of printed training and test shapes, and thereby can advance AM as a high-quality manufacturing paradigm.<br></div><div><br></div><div><b>2. Multiple Imputation and the Bootstrap for the Analysis of Big Data with Missingness</b></div><div><br></div><div>Inference can be a challenging task for Big Data. Two significant issues are that Big Data frequently exhibit complicated missing data patterns, and that the complex statistical models and machine learning algorithms typically used to analyze Big Data do not have convenient quantification of uncertainties for estimators. These two difficulties have previously been addressed using multiple imputation and the bootstrap, respectively. However, it is not clear how multiple imputation and bootstrap procedures can be effectively combined to perform statistical inferences on Big Data with missing values. We investigate a practical framework for the combination of multiple imputation and bootstrap methods. Our framework is based on two principles: distribution of multiple imputation and bootstrap calculations across parallel computational cores, and the quantification of sources of variability involved in bootstrap procedures that use subsampling techniques via random effects or hierarchical models. This framework effectively extends the scope of existing methods for multiple imputation and the bootstrap to a broad range of Big Data settings. We perform simulation studies for linear and logistic regression across Big Data settings with different rates of missingness to characterize the frequentist properties and computational efficiencies of the combinations of multiple imputation and the bootstrap. We further illustrate how effective combinations of multiple imputation and the bootstrap for Big Data analyses can be identified in practice by means of both the simulation studies and a case study on COVID infection status data. Ultimately, our investigation demonstrates how the flexible combination of multiple imputation and the bootstrap under our framework can enable valid statistical inferences in an effective manner for Big Data with missingness.<br></div>