This study investigates the process of index construction as a means of measuring a hypothetical construct that can typically not be measured by a single question or item in a survey study and applying it as a method of market segmentation. The availability of incidental secondary data that were gathered during 2009 provides a relevant quantitative basis to illustrate this process by constructing a commercial farming sophistication index for South Africa. A multi-step approach was followed for the construction of the commercial farming sophistication index, namely: (1) Selection of items and definition of variables that are most likely to be indicators of commercial farming sophistication; (2) combining of variables into an index; and (3) segmentation and index validation. Following the investigation and illustration of the process of index construction as a method of market segmentation, it was evident that this approach offers an appropriate and useful means of segmenting a market. Several factors contribute to the appeal of this approach. Amongst other, it contributes towards addressing important priorities in the area of future segmentation research, namely that of investigating the application of new base variables into segmentation models, as well as investigating new segmentation strategies. The approach also applies a creative process of combining several base variables into a single measure, namely that of an index variable. By offering classification rules based on characteristics that can easily be observed or elicited by asking a few key questions, new or potential buyers can be grouped by buying behaviour segment. Furthermore, the multi-step process that was employed has pragmatic appeal for researcher, and provides a systematic and structured multivariate approach to segmentation. It also facilitates replication of the process when conducting future studies. By using an index, it takes advantage of any intensity structure that may exist among attributes. This has the advantage that it places members of the market on a continuum that can lead to tracking members’ development paths as they progress towards higher levels on the index. Furthermore, illustration of the process has significant application value in other business-to-business markets, locally and internationally, where index variables can be constructed from both primary and secondary sources and used as a method of segmentation following a similar multi-step approach proposed in this study. Lastly, the outcome of this type of segmentation method offers researchers and marketing practitioners a procedure, in the form of an equation, to calculate index scores and provide rules to segment the market based on predefined intervals. Hence, the challenge to replicate segment formation across independent future studies is addressed. This process is considered an advantage over employing a technique such as cluster analysis, where the use of new data or changes to the clustering algorithm often leads to different segment solutions. / Thesis (PhD)--University of Pretoria, 2012. / Marketing Management / PhD / Unrestricted

The goal of this work was to design fast and simple methods for processing point-cloud-data of urban areas for virtual reality applications. For the visualization of methods, we developed a simple renderer written in C++ and HLSL. The renderer is based on DirectX 11. For point-cloud processing, we designed a method based on height-histograms for filtering ground points out of point cloud. We also proposed a parallel method for point cloud segmentation based on the region growing algorithm. The individual segments are then tested by simple rules to check if it is or it is not corresponding to a predefined object.

<div><p>Deep neural networks are essential in applications such as image categorization, natural language processing, autonomous driving, home automation, and robotics. Most of these applications require instantaneous processing of data and decision making. In general existing neural networks are computationally expensive, and hence they fail to perform in real-time. Models performing semantic segmentation are being extensively used in self-driving vehicles. Autonomous vehicles not only need segmented output, but also control system capable of processing segmented output and deciding actuator outputs such as speed and direction.</p> <p><br></p> <p>In this thesis we propose efficient neural network architectures with fewer operations and parameters as compared to current state-of-the-art algorithms. Our work mainly focuses on designing deep neural network architectures for semantic segmentation. First, we introduce few network modules and concepts which help in reducing model complexity. Later on, we show that in terms of accuracy our proposed networks perform better or at least at par with state-of-the-art neural networks. Apart from that, we also compare our networks' performance on edge devices such as Nvidia TX1. Lastly, we present a control system capable of predicting steering angle and speed of a vehicle based on the neural network output.</p></div>

The aim of this thesis is to build and evaluate how a word segmentation algorithm performs when extracting words from historical handwritten documents. Since historical documents often consist of background noise, the aim will also be to investigate whether applying a background removal algorithm will affect the final result or not. Three different types of historical handwritten documents are used to be able to compare the output when applying two different word segmentation algorithms. The result attained indicates that the background removal algorithm increases the accuracy obtained when using the word segmentation algorithm. The word segmentation algorithm developed successfully manages to extract a majority of the words while the obtained algorithm has difficulties for some documents. A conclusion made was that the type of document plays the key role in whether a poor result will be obtained or not. Hence, different algorithms may be needed rather than using one for all types of documents.

(Geo)demographic factors of consumers' behaviour Abstract This master's thesis deals with the issue of consumers' behaviour on theoretical level with empirical measuring of customers' satisfaction related to the selected mobile operator. The object of survey is the consumer and the subject matter comprises his behaviour in consumption, resp. his behaviour after purchase. The aim of the analysis is to conduct a test of relations between variables, to search for statistically significant differences between average values assigned to the overall satisfaction among particular groups of consumers divided using (geo) demographic and social variables, to describe segments created with reference to the level of satisfaction and to define such factors, which contribute towards the resultant value of satisfaction declared by significant means. The data analysis is based on selection survey conducted in the year 2008, using the method of individual, and standardised, assisted questioning. The processing of data is implemented, one- or multi-dimensional statistics - testing of average values, dispersion analysis, correlation, regression and factor analyses. That resulted in the finding that there was a relation between the overall satisfaction and gender, age, education, marital status, employment, territory. The most...

Segmenting brain MR scans could be highly benecial for diagnosing, treating and evaluating the progress of specic diseases. Up to this point, manual segmentation,performed by experts, is the conventional method in hospitals and clinical environments. Although manual segmentation is accurate, it is time consuming, expensive and might not be reliable. Many non-automatic and semi automatic methods have been proposed in the literature in order to segment MR brain images, but the levelof accuracy is not comparable with manual segmentation. The aim of this project is to implement and make a preliminary evaluation of a method based on machine learning technique for segmenting gray matter (GM),white matter (WM) and cerebrospinal uid (CSF) of brain MR scans using images available within the open MICCAI grand challenge (MRBrainS13).The proposed method employs supervised articial neural network based autocontext algorithm, exploiting intensity-based, spatial-based and shape model-basedlevel set segmentation results as features of the network. The obtained average results based on Dice similarity index were 97.73%, 95.37%, 82.76%, 88.47% and 84.78% for intracranial volume, brain (WM + GM), CSF, WM and GM respectively. This method achieved competitive results with considerably shorter required training time in MRBrainsS13 challenge.

Visual inspection of fuel assemblies is necessary to identify potential anomalies in their behaviour associated with their condition and their future usage. One of the possible find- ings are foreign objects caught on the fuel spacer grid which can disrupt the cladding of fuel rods during the operation. The goal of this thesis is to accurately segment the spacer grid from an image, which is a task dual to the foreign object detection, and therefore to automate visual inspection process in this area. We created new datasets covering typical problems appearing on the fuel assembly. To perform the segmentation, we em- ployed neural networks. We increased performance by data augmentation techniques and domain-specific output post-processing. We also measured the algorithm's performance by a newly introduced Line Distance metric, computing the size of the maximum un- certain area between the actual and the predicted transition between grids and rods. In the experiments, we found the best hyperparameters and reached very good results, outperforming our predecessor's algorithm by having three times lower Line Distance metric. 1

Die vorliegende Arbeit untersuchte die Durchführbarkeit und Genauigkeit der manuellen Segmentierung des Nervus facialis in hochauflösenden CT – Bildern im Vergleich mit einer semi – automatischen Segmentierung durch verschiedene Probanden. Die Untersuchung erfolgte für beide Segmentierformen Software gestützt. Zusätzlich wurde das subjektive Vertrauen der Probanden in die eigene Segmentierung abgefragt. Die beiden in dieser Arbeit untersuchten Methoden wiesen keinen signifikanten Unterschied für die Genauigkeit auf. Dabei betrug die mittlere quadratische Abweichung für die manuelle Segmentierform 0,53 mm und 0,42 mm für den semi-automatischen Ansatz. Die semi – automatische Segmentierung benötigte zudem mit einer mittleren Segmentierzeit/Datensatz von t = 94 sec deutlich weniger Zeit als die Probanden der manuellen Segmentierung mit einer mittleren Segmentierzeit/Datensatz von t = 219,5 sec. Anhand einer Korrelationsanalyse konnte festgestellt werden, dass bei der manuellen Segmentierung die Höhe der Abweichung (in mm) nicht mit der Segmentierzeit (in sec) korrelierte. Für den semi-automatischen Ansatz ergab sich auch lediglich für den zweiten Durchlauf eine Korrelation zwischen Höhe der Abweichung (in mm) und der Segmentierzeit (in sec) auf Signifikanzniveau (p < 0,01): Je mehr Zeit der Proband für seine Segmentierung in Anspruch nahm, desto höher fiel die Abweichung vom Referenzverlauf aus. Die Abfrage des subjektiven Segmentierungsvertrauens zeigte deutlich, dass sich die Probanden in ihrer Festlegung des Fazialisverlaufes sicher waren. Des Weiteren traten keine Unterschiede des Segmentierungsvertrauens zwischen den verschiedenen Durchläufen auf. Im Ergebnis zeigen die manuelle und die semi-automatische Segmentierung keine signifikanten Genauigkeitsunterschiede im Vergleich zur Referenz und können genutzt werden, um vor Eingriffen am Felsenbein eine sichere Identifikation des Nervus facialis zu liefern.

<p>Purpose - The goal of this study is to identify the current distinct market segments within the US agricultural credit lending market, predict segment membership based on readily available characteristics, and better understand farmer financing preferences. </p> <p><br></p> <p>Design/methodology/approach - A two stage clustering analysis was used to identify five distinct market segments. A multinomial logit regression was used to predict segment membership based on demographic and psychographic characteristics. </p> <p><br></p> <p>Findings - The segmentation analysis produced five distinct market segments. The identified segments are service, convenience, balance, price, and performance. </p> <p><br></p> <p>Practical implications - This information can aid credit lenders in segmenting the market and tailoring their sales approach to the different farmer segments. </p> <p><br></p> <p>Originality/value - This paper contributes to the literature in several ways. First, previous studies of farmer selection of lending institutions rely on supply side data (Brewer et al., 2019; Dodson & Koenig, 2004; Ifft and Fiechter, 2020). While these studies are useful in knowing how farmers may be segmented according to their choice set of particular lending institutions, what we cannot examine is why the farmer is choosing that choice set. Our study incorporates psychographic and buying preferences. Prior work has highlighted the trend away from demographics and socioeconomic characteristics towards psychographic characteristics as categories for customer segmentation (Sherrick et al., 1994). Secondly, as described above, much has changed in the agricultural lending markets concerning the lending institutions available to farmers and the technology that changes how farmers and lending institutions interact. Thus, this study updates the literature as farmers preferences may have changed due to the new market structure </p>

This thesis presents a case study on the application of deep learning methods for the dense prediction of land cover types in oblique ground-based photography. While deep learning approaches are widely used in land cover classification of remote-sensing data (i.e., aerial and satellite orthoimagery) for change detection analysis, dense classification of oblique landscape imagery used in repeat photography remains undeveloped. A performance evaluation was carried out to test two state-of the-art architectures, U-net and Deeplabv3+, as well as a fully-connected conditional random fields model used to boost segmentation accuracy. The evaluation focuses on the use of a novel threshold-based data augmentation technique, and three multi-loss functions selected to mitigate class imbalance and input noise. The dataset used for this study was sampled from the Mountain Legacy Project (MLP) collection, comprised of high-resolution historic (grayscale) survey photographs of Canada’s Western mountains captured from the 1880s through the 1950s and their corresponding modern (colour) repeat images. Land cover segmentations manually created by MLP researchers were used as ground truth labels. Experimental results showed top overall F1 scores of 0.841 for historic models, and 0.909 for repeat models. Data augmentation showed modest improvements to overall accuracy (+3.0% historic / +1.0% repeat), but much larger gains for under-represented classes. / Graduate