Map matching is a widely used technology for assigning tracks recorded by Global Navigation Satellite Systems (GNSS) to existing road networks. Due to the measurement uncertainty of GNSS positions, the biggest challenge is to map them accurately. To develop and verify suitable algorithms for map matching, ground truth, i.e., the traveled routes in the road network, is required as a reference. However, GNSS recorded tracks naturally lack the ground truth routes. Providing this data is time-consuming and costly in these cases, as it requires manual correction of the routes based on human memorization. This is not practical on a large scale, e.g., with floating car data (FCD). This is why there exist only a few isolated ground truth data sets that were created in this way for map matching. To close this gap, we introduce and evaluate in this work a new open source tool-chain for artificially generating large amounts of simulated ground truth routes for map matching. Based on these routes, we generate simulated FCD and we apply comparably authentic and parameterizable artificial GNSS noise with varying noise characteristics. The generated data allows to thoroughly evaluate and improve the performance of existing map matching algorithms and facilitates in future research the development of novel algorithms based on sufficiently large and diverse labeled data. In this work, we evaluate different scenarios of varying noise characteristics of our artificially generated ground truth data to compare the robustness, individual strengths, and weaknesses of existing open source map matching implementations. Our new approach of artificially generating ground truth data for map matching addresses the existing lack of sufficient available reference data for ongoing map matching research.

There is a huge demand for trying on clothing at home. Recent methods to capture your figure mostly work in a 2d image plane and despite recent improvements of available technology the simulation of clothing is still not satisfactory. Especially the rendering of different clothing sizes is still a major challenge, which is only addressed by COTTON [1]. We propose an improvement to this approach by adding more control over the image generation process. For this we employ a special type of conditional diffusion model, namely ControlNet, and take keypoints of the fashion as conditional input. (This work is a part of the project (M4-SKI) has been supported and funded by the European Regional Development Fund (ERDF)).

In this paper we present a stable-diffusion based zero-shot approach to realistically transform the image of a

human body into a more fit version of that depicted person. Therefore we combine a modified stable diffusion

model with inpainting techniques and incorporated constraints. We introduce a prototype which allows users to

upload a photo and visualize a more fit version of themselves. We evaluated our approach in various experiments

and focused on the applicability and effectiveness of these techniques, with attention to gender-specific results.

This work contributes to the fields of computer vision and generative AI by demonstrating practical applications

and identifying areas for improvement in realistic body transformation visualizations. (This work is a part of the project (M4-SKI) has been supported and funded by the European Regional Development Fund (ERDF)).

The simulation of clothing for a virtual try on is still a challenging task, especially if the customer wants to use state of the art technology. To address this, we employ a 2D plane to process customer images. Specifically, we utilize a neural network, notably an autoencoder, to render so called fashion landmarks. As input we use human keypoints that represent the model poses and our fashion landmarks of the clothing from stock photos to generate fashion landmarks in the desired pose. These can be utilized by additional algorithms to adapt the clothing length or width.

This project has been funded by the European Regional Development Fund (EFRE).

In the midst of the growth in fashion retail industry, fitting of clothing remotely can be utilized to reduce return rates and unnecessary shipping. Due to recent advancements in Generative AI, models for generating a virtual try-on experience have also been able to develop further. This paper evaluates different models and techniques which are built for the application of virtual try-on. Additionally, this paper discusses about different models - ClothFlow model, Contextual-VTON model or FitGAN along with a deep insight into algorithms and methodologies namely Residual Networks, U-Net and Generative Adversarial Networks. Additionally, the paper compares these state-of-the-art models against the various evaluation metrics such as Frechet Inception Distance (FID), Inception Score (IS), Structural Similarity index (SSIM). It also explains, how the generated image outputs vary based on the changes made in hyper-parameters such as learning rate, batch size etc., and how these changes impact on the generated output. 

This project has been funded by the European Regional Development Fund (EFRE).

Map matching is a widely used technology for mapping tracks to road networks. Typically, tracks are recorded using publicly available Global Navigation Satellite Systems, and road networks are derived from the publicly available OpenStreetMap project. The challenge lies in resolving the discrepancies between the spatial location of the tracks and the underlying road network of the map. Map matching is a combination of defined models, algorithms, and metrics for resolving these differences that result from measurement and map errors. The goal is to find routes within the road network that best represent the given tracks. These matches allow further analysis since they are freed from the noise of the original track, they accurately overlap with the road network, and they are corrected for impossible detours and gaps that were present in the original track. Given the ongoing need for map matching in mobility research, in this work, we present a novel map matching method based on Markov decision processes with Reinforcement Learning algorithms. We introduce the new Candidate Adoption feature, which allows our model to dynamically resolve outliers and noise clusters. We also incorporate an improved Trajectory Simplification preprocessing algorithm for further improving our performance. In addition, we introduce a new map matching metric that evaluates direction changes in the routes, which effectively reduces detours and round trips in the results. We provide our map matching implementation as Open Source Software (OSS) and compare our new approach with multiple existing OSS solutions on several public data sets. Our novel method is more robust to noise and outliers than existing methods and it outperforms them in terms of accuracy and computational speed.

Um den zunehmenden Anforderungen an die Beherrschung digitaler Techniken und an die Fähigkeit zur interdisziplinären Zusammenarbeit an Studierende aller Fach­richtungen zu begegnen wurde das interdisziplinäre Lehrformat Digitalisie­rungs­­kollegs für Studierende entwickelt. Das in vielen Fachbereichen ausbaufähige Angebot von Digitalthemen in der Hoch­schullehre wird hiermit dauerhaft erwei­tert. Ein Digitalisierungskolleg besteht aus einer Vorlesungs­reihe mit an­gren­zen­dem Seminar, in denen Studierende interdisziplinäre Lösungen für Fragen der digitalen Transformation entwickeln. Geleitet werden sie von etablierten Wissen­schaftlerinnen und Wissenschaftlern, aktiv betreut und ausge­staltet von ein bis zwei Coaches. Kernelement sowohl des Kollegs als auch der einzelnen Projekte ist die Interdisziplinarität. Eine*r der beteiligten Projekt­leiter*innen hat einen direkten Bezug zur Technik und kommt aus der Informatik, der Wirtschafts­informatik, der Elektro­technik oder vergleichbaren Disziplinen. Zielgruppe der Projekte sind Studierende verschiedener Disziplinen im Masterstudium oder in den letzten Semestern eines Bachelorstudiums. Durch die Teilnahme er­werben auch Studie­ren­de aus digitali­­sie­rungs­fernen Fächern frühzeitig umfang­reiche IT-Kenntnisse. Als Be­gleiteffekt der um­fang­reichen Vernetzung zwischen den Digi­talisierungs­kollegs (Studierende, Coaches und Projektleitende) entsteht bereits zu Beginn einer wissenschaftlichen Karriere eine große digitale Community. Alle Teil­nehmenden lernen frühzeitig die interdisziplinäre Zusammenarbeit und ver­bessern erheblich ihre Karriere­chancen inner­­halb und außer­halb der Wissenschaft.

EEG classification is a promising approach to facilitate the life of handicapped people and to generate future human-computer-interfaces. In this paper we want to compare the effectiveness of current state of the art deep learning techniques for EEG classification. Therefore, we applied different approaches on various datasets and did a crosscomparison of the results in order to get more knowledge on the generalization capabilities. Additionally, we created a new EEG dataset and made it available for further research.

To provide robots for a wide range of users, there needs to be an easy and intuitive way to program them. This issue is addressed by the robot programming by demonstration or imitation learning paradigm, where the user demonstrates the task to the robot by teleoperation. Although single-shot approaches could save a lot of time and effort, they are still a niche due to some drawbacks, like ambiguities in selecting the relevant features.In this work we try to enhance a single shot programming by demonstration approach on sub-symbolic level by extending it to a multi modal input. While most approaches mainly focus on the trajectories and visual detection of objects, we combine speech and kinestethic teaching in order to resolve ambiguities and to rise the level of transferred information.

Überall hält heutzutage der Trend der Digitalisierung Einzug und verändert Ansprüche und Zielsetzungen. Die Organisationen der Wohlfahrts- und Sozialwirtschaft müssen mit ihren Ressourcen, vor allem mit ihren Mitarbeitenden dieser Entwicklung Rechnung tragen. Mehr denn je wird deutlich, dass die Digitalisierung keinen Zukunftstrend mehr darstellt, sondern bereits die Gegenwart beherrscht. Sie entfaltet heute schon Wirkung in Unternehmen und Organisationen bzw. in der Wohlfahrts- und Sozialwirtschaft und weit darüber hinaus. Die vielfältigen und anspruchsvollen Herausforderungen sind offensichtlich. Die vorliegende Publikation stellt die breite und umfassende, gleichsam systematische Aufarbeitung dieses komplexen Themenfeldes in den Vordergrund – und dies verknüpft mit erfahrungsbasierten Orientierungen.

Das Schlagwort „Big Data“ verspricht die Gewinnung relevanter Informationen durch die automatisierte Erfassung und Analyse sehr großer Datenmengen für unterschiedliche Anwendungsbereiche. Damit lassen sich praktisch alle wesentlichen Informationen zur Bewertung einer komplexen Situation erfassen. Big-Data-Technologien können dann mit Hilfe geeigneter Indikatoren Situationen in Echtzeit bewerten und genaue Prognosen ermöglichen. Wesentlich für diese Technologien ist die Verarbeitung großer Datenmengen unter engen zeitlichen Rahmenbedingungen, um die Aktualität der Ergebnisse sicherzustellen. Existierende betriebliche Informationssysteme auf der Basis relationaler Datenbankmanagementsysteme erreichen dabei ihre Grenzen und können in der Regel die geforderten Antwortzeiten nicht mehr erfüllen. Neuere Datenbanktechnologien versprechen hier einen deutlichen Effizienzgewinn, so dass auch sehr große Datenmengen im Rahmen interaktiver Anwendungen verarbeitet und analysiert werden können. Dieses Kapitel gibt anhand eines Beispiels zur Erfassung und Auswertung von Daten einer Betriebsdatenerfassung einen Überblick über diese Technologien.

Hybrid index structures support access to heterogeneous data types in multiple columns. Several experiments confirm the improved efficiency of these hybrid access structures. Yet, very little is known about the worst case time and space complexity of them. This paper aims to close this gap by introducing a theoretical framework supporting the analysis of hybrid index structures. This framework then is used to derive the constraints for an access structure which is both time and space efficient. An access structure based on a B+-Tree augmented with bit lists representing sets of terms from texts is the outcome of the analysis which is then validated experimentally together with a hybrid R-Tree variant to show a logarithmic search time complexity.