Parsing floorplans have been a problem in automatic document analysis for long and have up until recent years been approached with algorithmic methods. With the rise of convolutional neural networks (CNN), this problem too has seen an upswing in performance. In this thesis the task is to recover, as accurately as possible, spatial and geometric information from floorplans. This project builds around instance segmentation models like Cascade Mask R-CNN to extract the bulk of information from a floorplan image. To complement the segmentation, a new style of using keypoint-CNN is presented to find precise locations of corners. These are then combined in a post-processing step to give the resulting segmentation. The resulting segmentation scores exceed the current baseline of the CubiCasa5k floorplan dataset with a mean IoU of 72.7% compared to 57.5%. Further, the mean IoU for individual classes is also improved for almost every class. It is also shown that Cascade Mask R-CNN is better suited than Mask R-CNN for this task.

In this century and the last, serious efforts have been made to digitize the content housed by libraries across the world. In order to open up these volumes to content-based information retrieval, independent elements such as headlines, body text, bylines, images and captions ideally need to be connected semantically as article-level units. To query on facets such as author, section, content type or other metadata, further processing of these documents is required. Even though humans have shown exceptional ability to segment different types of elements into related components, even in languages foreign to them, this task has proven difficult for computers. The challenge of semantic segmentation in newspapers lies in the diversity of the medium: Newspapers have vastly different layouts, covering diverse content, from news articles to ads to weather reports. State-of-the-art object detection and segmentation models have been trained to detect and segment real-world objects. It is not clear whether these architectures can perform equally well when applied to scanned images of printed text. In the domain of newspapers, in addition to the images themselves, we have access to textual information through Optical Character Recognition. The recent progress made in the field of instance segmentation of real-world objects using deep learning techniques begs the question: Can the same methodology be applied in the domain of newspaper articles? In this thesis we investigate one possible approach to encode the textual signal into the image in an attempt to improve performance. Based on newspapers from the National Library of Sweden, we investigate the predictive power of visual and textual features and their capacity to generalize across different typographic designs. Results show impressive mean Average Precision scores (>0:9) for test sets sampled from the same newspaper designs as the training data when using only the image modality. / I detta och det förra århundradet har kraftiga åtaganden gjorts för att digitalisera traditionellt medieinnehåll som tidigare endast tryckts i pappersformat. För att kunna stödja sökningar och fasetter i detta innehåll krävs bearbetning påsemantisk nivå, det vill säga att innehållet styckas upp påartikelnivå, istället för per sida. Trots att människor har lätt att dela upp innehåll påsemantisk nivå, även påett främmande språk, fortsätter arbetet för automatisering av denna uppgift. Utmaningen i att segmentera nyhetsartiklar återfinns i mångfalden av utseende och format. Innehållet är även detta mångfaldigt, där man återfinner allt ifrån faktamässiga artiklar, till debatter, listor av fakta och upplysningar, reklam och väder bland annat. Stora framsteg har gjorts inom djupinlärning just för objektdetektering och semantisk segmentering bara de senaste årtiondet. Frågan vi ställer oss är: Kan samma metodik appliceras inom domänen nyhetsartiklar? Dessa modeller är skapta för att klassificera världsliga ting. I denna domän har vi tillgång till texten och dess koordinater via en potentiellt bristfällig optisk teckenigenkänning. Vi undersöker ett sätt att utnyttja denna textinformation i ett försök att förbättra resultatet i denna specifika domän. Baserat pådata från Kungliga Biblioteket undersöker vi hur väl denna metod lämpar sig för uppstyckandet av innehåll i tidningar längsmed tidsperioder där designen förändrar sig markant. Resultaten visar att Mask R-CNN lämpar sig väl för användning inom domänen nyhetsartikelsegmentering, även utan texten som input till modellen.

This thesis investigates the performance of Mask R-CNN when utilizing transfer learning on a small dataset. The aim was to instance segment eyeglass lenses as accurately as possible from self-portrait images. Five different models were trained, where the key difference was the types of eyeglasses the models were trained on. The eyeglasses were grouped into three types, fully rimmed, semi-rimless, and rimless glasses. 1550 images were used for training, validation, and testing. The model's performances were evaluated using TensorBoard training data and mean Intersection over Union scores (mIoU). No major differences in performance were found in four of the models, which grouped all three types of glasses into one class. Their mIoU scores range from 0.913 to 0.94 whereas the model with one class for each group of glasses, performed worse, with a mIoU of 0.85. The thesis revealed that one can achieve great instance segmentation results using a limited dataset when taking advantage of transfer learning. / Denna uppsats undersöker prestandan för Mask R-CNN vid användning av överföringsinlärning på en liten datamängd. Syftet med arbetet var att segmentera glasögonlinser så exakt som möjligt från självporträttbilder. Fem olika modeller tränades, där den viktigaste skillnaden var de typer av glasögon som modellerna tränades på. Glasögonen delades in i 3 typer, helbåge, halvbåge och båglösa. Totalt samlades 1550 träningsbilder in, dessa annoterades och användes för att träna modellerna.  Modellens prestanda utvärderades med TensorBoard träningsdata samt genomsnittlig Intersection over Union (IoU). Inga större skillnader i prestanda hittades mellan modellerna som endast tränades på en klass av glasögon. Deras genomsnittliga IoU varierar mellan 0,913 och 0,94. Modellen där varje glasögonkategori representerades som en unik klass, presterade sämre med en genomsnittlig IoU på 0,85. Resultatet av uppsatsen påvisar att goda instanssegmenteringsresultat går att uppnå med hjälp av en begränsad datamängd om överföringsinlärning används.

In modern industrial environments, human-robot collaboration is a trend in automation to improve performance and productivity. Instead of isolating robot from human to guarantee safety, collaborative robotics allows human and robot working in the same area at the same time. New hazards and risks, such as the collision between robot and human, arise in this situation. Safety analysis is necessary to protect both human and robot when using a collaborative robot.To perform safety analysis, robots need to perceive the surrounding environment in realtime. This surrounding environment is perceived and stored in the form of scene graph, which is a direct graph with semantic representation of the environment, the relationship between the detected objects and properties of these objects. In order to generate the scene graph, a simulated warehouse is used: robots and humans work in a common area for transferring products between shelves and conveyor belts. Each robot generates its own scene graph from the attached camera sensor. In the graph, each detected object is represented by a node and edges are used to denote the relationship among the identified objects. The graph node includes values like velocity, bounding box sizes, orientation, distance and directions between the object and the robot.We generate scene graph in a simulated warehouse scenario with the frequency of 7 Hz and present a study of Mask R-CNN based on the qualitative comparison. Mask R-CNN is a method for object instance segmentation to get the properties of the objects. It uses ResNetFPN for feature extraction and adds a branch to Faster R-CNN for predicting segmentation mask for each object. And its results outperform almost all existing, single-model entries on instance segmentation and bounding-box object detection. With the help of this method, the boundaries of the detected object are extracted from the camera images. We initialize Mask R-CNN model using three different types of weights: COCO pre-trained weight, ImageNet pre-trained weight and random weight, and the results of these three different weights are compared w.r.t. precision and recall.Results showed that Mask R-CNN is also suitable for simulated environments and can meet requirements in both detection precision and speed. Moreover, the model trained used the COCO pre-trained weight outperformed the model with ImageNet and randomly assigned initial weights. The calculated Mean Average Precision (mAP) value for validation dataset reaches 0.949 with COCO pre-trained weights and execution speed of 11.35 fps. / I modern industriella miljöer, för att förbättra prestanda och produktivitet i automatisering är human-robot samarbete en trend. Istället för att isolera roboten från människan för att garantera säkerheten, möjliggör samarbets robotar att man och robot arbetar i samma område samtidigt. Nya risker, såsom kollisionen mellan robot och människa, uppstår i denna situation. Säkerhetsanalys är nödvändig för att skydda både människa och robot när man använder en samarbets robot.För att utföra säkerhetsanalys måste robotar uppfatta omgivningen i realtid. Denna omgivande miljö uppfattas och lagras i form av scen graf, som är ett direkt diagram med semantisk representation av miljön, samt förhållandet mellan de detekterade objekten och egenskaperna hos dessa objekt. För att skapa scen grafen används ett simulerat lager: robotar och människor arbetar i ett gemensamt område för överföring av produkter mellan hyllor och transportband. Varje robot genererar sin egen scen grafik från den medföljande kamerasensorn. I diagrammet presenteras varje detekterat objekt av en nod och kanterna används för att beteckna förhållandet mellan de identifierade objekten. Diagram noden innehåller värden som hastighet, gränsvärde, orientering, avstånd och riktningar mellan objektet och roboten.Vi genererar scen graf i ett simulerat lager scenario med frekvensen 7 Hz och presenterar en studie av Mask R-CNN baserat på den kvalitativa jämförelsen. Mask R-CNN är ett sätt att segmentera objekt exempel för att få objektens egenskaper. Det använder ResNetFPN för funktion extraktion och lägger till en gren till Snabbare R-CNN för att förutsäga segmenterings mask för varje objekt. Och dess resultat överträffar nästan alla befintliga, enkel modell poster, till exempel segmentering och avgränsning av objektiv detektering. Med hjälp av denna metod extraheras kanterna för det detekterade objektet från kamerabilderna. Vi initierar Mask R-CNN-modellen med tre olika typer av vikter: COCO-utbildade vikter, ImageNet-tränade vikter och slumpmässiga vikter, och resultaten av dessa tre olika vikter jämförs med avseende på precision och återkallelse.Resultaten visade att Mask R-CNN också är lämplig för simulerade miljöer och kan uppfylla kraven i både detekterings precision och hastighet. Dessutom använde den utbildade modellen de COCO-tränade vikterna överträffat modellen med slumpmässigt tilldelade initial vikter. Det beräknade medelvärdet för precision (mAP) för validerings dataset når 0.949 med COCO-pre-utbildade vikter och körhastighet på 11.35 fps.

No description available.

Thesis (PhD (Mathematical Sciences))--University of Stellenbosch, 2007. / We study refinable functions where the dilation factor is not always assumed to be 2. In our investigation, the role of convolutions and refinable step functions is emphasized as a framework for understanding various previously published results. Of particular importance is a class of polynomial factors, which was first introduced for dilation factor 2 by Berg and Plonka and which we generalise to general integer dilation factors. We obtain results on the existence of refinable functions corresponding to certain reduced masks which generalise similar results for dilation factor 2, where our proofs do not rely on Fourier methods as those in the existing literature do. We also consider subdivision for general integer dilation factors. In this regard, we extend previous results of De Villiers on refinable function existence and subdivision convergence in the case of positive masks from dilation factor 2 to general integer dilation factors. We also obtain results on the preservation of subdivision convergence, as well as on the convergence rate of the subdivision algorithm, when generalised Berg-Plonka polynomial factors are added to the mask symbol. We obtain sufficient conditions for the occurrence of polynomial sections in refinable functions and construct families of related refinable functions. We also obtain results on the regularity of a refinable function in terms of the mask symbol factorisation. In this regard, we obtain much more general sufficient conditions than those previously published, while for dilation factor 2, we obtain a characterisation of refinable functions with a given number of continuous derivatives. We also study the phenomenon of subsequence convergence in subdivision, which explains some of the behaviour that we observed in non-convergent subdivision processes during numerical experimentation. Here we are able to establish different sets of sufficient conditions for this to occur, with some results similar to standard subdivision convergence, e.g. that the limit function is refinable. These results provide generalisations of the corresponding results for subdivision, since subsequence convergence is a generalisation of subdivision convergence. The nature of this phenomenon is such that the standard subdivision algorithm can be extended in a trivial manner to allow it to work in instances where it previously failed. Lastly, we show how, for masks of length 3, explicit formulas for refinable functions can be used to calculate the exact values of the refinable function at rational points. Various examples with accompanying figures are given throughout the text to illustrate our results.

The traditional approach to imaging system design places the sole burden of image formation on optical components. In contrast, a computational imaging system relies on a combination of optics and post-processing to produce the final image and/or output measurement. Therefore, the joint-optimization (JO) of the optical and the post-processing degrees of freedom plays a critical role in the design of computational imaging systems. The JO framework also allows us to incorporate task-specific performance measures to optimize an imaging system for a specific task. In this dissertation, we consider the design of computational imaging systems within a JO framework for two separate tasks: object reconstruction and iris-recognition. The goal of these design studies is to optimize the imaging system to overcome the performance degradations introduced by under-sampled image measurements. Within the JO framework, we engineer the optical point spread function (PSF) of the imager, representing the optical degrees of freedom, in conjunction with the post-processing algorithm parameters to maximize the task performance. For the object reconstruction task, the optimized imaging system achieves a 50% improvement in resolution and nearly 20% lower reconstruction root-mean-square-error (RMSE ) as compared to the un-optimized imaging system. For the iris-recognition task, the optimized imaging system achieves a 33% improvement in false rejection ratio (FRR) for a fixed alarm ratio (FAR) relative to the conventional imaging system. The effect of the performance measures like resolution, RMSE, FRR, and FAR on the optimal design highlights the crucial role of task-specific design metrics in the JO framework. We introduce a fundamental measure of task-specific performance known as task-specific information (TSI), an information-theoretic measure that quantifies the information content of an image measurement relevant to a specific task. A variety of source-models are derived to illustrate the application of a TSI-based analysis to conventional and compressive imaging (CI) systems for various tasks such as target detection and classification. A TSI-based design and optimization framework is also developed and applied to the design of CI systems for the task of target detection, it yields a six-fold performance improvement over the conventional imaging system at low signal-to-noise ratios.

A common problem following general anesthesia is postoperative sore throat (POST) and postoperative hoarseness (PH). Symptoms directly correlated with less satisfaction according to the patients. The overall aim of this thesis was to describe patients' postoperative sore throat and hoarseness after general anesthesia with endotracheal intubation or laryngeal mask airway. As well as to investigate the risk factors that are associated with the symptoms, and to test methods that may prevent sore throat and hoarseness after a general anaesthetics. A total of 889 patients are included in the four studies. Incidence of POST varied from 21% up to 52 % depending on endotracheal tube (ETT) size in women (I-IV) and in men was the incidence 32-38% (III-IV). There were no gender difference in POST in study III and IV. The overall incidence of PH varied from 42- 59% (I-IV) in all patients, with no gender differences (III-IV). Following a laryngeal mask airway (LMA) 19% of the patients had POST and 33% of the patients reported PH. Patients with POST do seem to be able to localize their pain in the throat (IV). Different risk factors are shown to contribute to both POST and PH in men and women (II-III). To intubate with a smaller ETT size, 6.0 vs. 7.0 decreased POST in women in the early postoperative period as well as their discomfort from their POST (I). Only 6% of men who needed a laryngeal mask airway had POST compared to 26% of women. The symptoms are more discomforting after an ETT vs. an LMA up to 24 hours (IV). More patients have sore throat and hoarseness in the early postoperative period, but the symptoms can remain up to almost 5 days postoperatively (I, IV). In summary, sore throat and hoarseness following general anesthesia, affects many patients postoperatively. To intubate women with endotracheal size 6.0 decreases both sore throat and hoarseness postoperatively. Women are more likely than men to have a sore throat when a laryngeal mask airway is used.

It is the purpose of this thesis to explore the different principles of mask performance in modern theatre and the unique relationship it allows the actor to develop with an audience. The author uses exercises from her training experience with different mask artists such as Teatro Punto, Familie Floz, Torbjorn Alstrom, and Marcello Bartoli. These exercises document a process of training the body in preparation for the mask, how to perform with mask, and how to connect with an audience. The mask is used as a tool to discover an engagement of work that activates both body and imagination of the performer as well as the observer.

O trabalho apresenta algumas das iniciativas para a retomada da máscara no teatro moderno. No amplo quadro da primeira metade do século XX, escolhemos quatro principais artistas que abordaram esse objeto: Gordon Craig, Meyerhold, Jacques Copeau e Jacques Lecoq. Nestes a máscara não é apenas mais um elemento colocado em cena, mas também um princípio organizador da teatralidade que almejam. Através de seus trabalhos analisamos algumas características da máscara que podem ser reelaboradas no teatro contemporâneo. Em Gordon Craig, abordamos as primeiras proposições dentro do teatro simbolista, além do panorama da máscara no início do século XX; através de Meyerhold, pudemos traçar os elementos grotescos contidos na máscara; com Jacques Copeau, questionamos os conceitos de caráter e de tipo e, através da pedagogia de Jacques Lecoq, pudemos analisar o trabalho da máscara como signo representativo. A partir dos elementos abordados, apontamos uma proposição de trabalho com a máscara que procura aproximá-la do conceito de figura. / This study presents some of the initiatives to reinstate the mask in modern theatre. In the vast spectrum of the first half of the Twentieth Century, I chose four key artists who elaborated this object: Gordon Craig, Meyerhold, Jacques Copeau and Jacques Lecoq. In the work of these artists, the mask is not one scenic element among others, but becomes the organizing keystone for the theatricality that they seek. Through their creations we can analyze some of the mask´s features that are being reelaborated in contemporary theatre. In Gordon Craig, we examine initial propositions within the symbolist theatre, placing this in a panorama of the mask at the beginning of the Twentieth Century; with Meyerhold, one can trace the grotesque elements contained in the mask; in Jacques Copeau, the focus moves to the concepts of character and type, and finally, in Jacques Lecoqs pedagogy, one can probe the mask as representational sign. From the foregoing elements, we outline an operational proposal which aims to approximate the mask to the concept of figure.