006 Spezielle Computerverfahren
Filtern
Dokumenttyp
Sprache
- Englisch (17) (entfernen)
Schlagworte
- Virtual Reality (3)
- Virtuelle Realität (3)
- virtual reality (3)
- Augmented Reality (2)
- Extended Reality (2)
- Social VR (2)
- 3D model generation (1)
- Autonomer Roboter (1)
- Bildverarbeitung (1)
- Deep learning (1)
Institut
- Institut für Informatik (12)
- Institut Mensch - Computer - Medien (2)
- Institut für Psychologie (2)
- Graduate School of Life Sciences (1)
- Graduate School of Science and Technology (1)
- Institut für Geographie und Geologie (1)
- Institut für Philosophie (1)
- Klinik und Poliklinik für Mund-, Kiefer- und Plastische Gesichtschirurgie (1)
- Pathologisches Institut (1)
Sonstige beteiligte Institutionen
This paper deals with the effect of exploiting background knowledge for improving an OMR (Optical Music Recognition) deep learning pipeline for transcribing medieval, monophonic, handwritten music from the 12th–14th century, whose usage has been neglected in the literature. Various types of background knowledge about overlapping notes and text, clefs, graphical connections (neumes) and their implications on the position in staff of the notes were used and evaluated. Moreover, the effect of different encoder/decoder architectures and of different datasets for training a mixed model and for document-specific fine-tuning based on an extended OMR pipeline with an additional post-processing step were evaluated. The use of background models improves all metrics and in particular the melody accuracy rate (mAR), which is based on the insert, delete and replace operations necessary to convert the generated melody into the correct melody. When using a mixed model and evaluating on a different dataset, our best model achieves without fine-tuning and without post-processing a mAR of 90.4%, which is raised by nearly 30% to 93.2% mAR using background knowledge. With additional fine-tuning, the contribution of post-processing is even greater: the basic mAR of 90.5% is raised by more than 50% to 95.8% mAR.
Background: Oro-antral communication (OAC) is a common complication following the extraction of upper molar teeth. The Archer and the Root Sinus (RS) systems can be used to classify impacted teeth in panoramic radiographs. The Archer classes B-D and the Root Sinus classes III, IV have been associated with an increased risk of OAC following tooth extraction in the upper molar region. In our previous study, we found that panoramic radiographs are not reliable for predicting OAC. This study aimed to (1) determine the feasibility of automating the classification (Archer/RS classes) of impacted teeth from panoramic radiographs, (2) determine the distribution of OAC stratified by classification system classes for the purposes of decision tree construction, and (3) determine the feasibility of automating the prediction of OAC utilizing the mentioned classification systems. Methods: We utilized multiple supervised pre-trained machine learning models (VGG16, ResNet50, Inceptionv3, EfficientNet, MobileNetV2), one custom-made convolutional neural network (CNN) model, and a Bag of Visual Words (BoVW) technique to evaluate the performance to predict the clinical classification systems RS and Archer from panoramic radiographs (Aim 1). We then used Chi-square Automatic Interaction Detectors (CHAID) to determine the distribution of OAC stratified by the Archer/RS classes to introduce a decision tree for simple use in clinics (Aim 2). Lastly, we tested the ability of a multilayer perceptron artificial neural network (MLP) and a radial basis function neural network (RBNN) to predict OAC based on the high-risk classes RS III, IV, and Archer B-D (Aim 3). Results: We achieved accuracies of up to 0.771 for EfficientNet and MobileNetV2 when examining the Archer classification. For the AUC, we obtained values of up to 0.902 for our custom-made CNN. In comparison, the detection of the RS classification achieved accuracies of up to 0.792 for the BoVW and an AUC of up to 0.716 for our custom-made CNN. Overall, the Archer classification was detected more reliably than the RS classification when considering all algorithms. CHAID predicted 77.4% correctness for the Archer classification and 81.4% for the RS classification. MLP (AUC: 0.590) and RBNN (AUC: 0.590) for the Archer classification as well as MLP 0.638) and RBNN (0.630) for the RS classification did not show sufficient predictive capability for OAC. Conclusions: The results reveal that impacted teeth can be classified using panoramic radiographs (best AUC: 0.902), and the classification systems can be stratified according to their relationship to OAC (81.4% correct for RS classification). However, the Archer and RS classes did not achieve satisfactory AUCs for predicting OAC (best AUC: 0.638). Additional research is needed to validate the results externally and to develop a reliable risk stratification tool based on the present findings.
In this work, a novel method for estimating the relative pose of a known object is presented, which relies on an application-specific data fusion process. A PMD-sensor in conjunction with a CCD-sensor is used to perform the pose estimation. Furthermore, the work provides a method for extending the measurement range of the PMD sensor along with the necessary calibration methodology. Finally, extensive measurements on a very accurate Rendezvous and Docking testbed are made to evaluate the performance, what includes a detailed discussion of lighting conditions.
Obesity is a serious disease that can affect both physical and psychological well-being. Due to weight stigmatization, many affected individuals suffer from body image disturbances whereby they perceive their body in a distorted way, evaluate it negatively, or neglect it. Beyond established interventions such as mirror exposure, recent advancements aim to complement body image treatments by the embodiment of visually altered virtual bodies in virtual reality (VR). We present a high-fidelity prototype of an advanced VR system that allows users to embody a rapidly generated personalized, photorealistic avatar and to realistically modulate its body weight in real-time within a carefully designed virtual environment. In a formative multi-method approach, a total of 12 participants rated the general user experience (UX) of our system during body scan and VR experience using semi-structured qualitative interviews and multiple quantitative UX measures. Using body weight modification tasks, we further compared three different interaction methods for real-time body weight modification and measured our system’s impact on the body image relevant measures body awareness and body weight perception. From the feedback received, demonstrating an already solid UX of our overall system and providing constructive input for further improvement, we derived a set of design guidelines to guide future development and evaluation processes of systems supporting body image interventions.
Software frameworks for Realtime Interactive Systems (RIS), e.g., in the areas of Virtual, Augmented, and Mixed Reality (VR, AR, and MR) or computer games, facilitate a multitude of functionalities by coupling diverse software modules. In this context, no uniform methodology for coupling these modules does exist; instead various purpose-built solutions have been proposed. As a consequence, important software qualities, such as maintainability, reusability, and adaptability, are impeded.
Many modern systems provide additional support for the integration of Artificial Intelligence (AI) methods to create so called intelligent virtual environments. These methods exacerbate the above-mentioned problem of coupling software modules in the thus created Intelligent Realtime Interactive Systems (IRIS) even more. This, on the one hand, is due to the commonly applied specialized data structures and asynchronous execution schemes, and the requirement for high consistency regarding content-wise coupled but functionally decoupled forms of data representation on the other.
This work proposes an approach to decoupling software modules in IRIS, which is based on the abstraction of architecture elements using a semantic Knowledge Representation Layer (KRL). The layer facilitates decoupling the required modules, provides a means for ensuring interface compatibility and consistency, and in the end constitutes an interface for symbolic AI methods.
Social robots in applied settings: a long-term study on adaptive robotic tutors in higher education
(2022)
Learning in higher education scenarios requires self-directed learning and the challenging task of self-motivation while individual support is rare. The integration of social robots to support learners has already shown promise to benefit the learning process in this area. In this paper, we focus on the applicability of an adaptive robotic tutor in a university setting. To this end, we conducted a long-term field study implementing an adaptive robotic tutor to support students with exam preparation over three sessions during one semester. In a mixed design, we compared the effect of an adaptive tutor to a control condition across all learning sessions. With the aim to benefit not only motivation but also academic success and the learning experience in general, we draw from research in adaptive tutoring, social robots in education, as well as our own prior work in this field. Our results show that opting in for the robotic tutoring is beneficial for students. We found significant subjective knowledge gain and increases in intrinsic motivation regarding the content of the course in general. Finally, participation resulted in a significantly better exam grade compared to students not participating. However, the extended adaptivity of the robotic tutor in the experimental condition did not seem to enhance learning, as we found no significant differences compared to a non-adaptive version of the robot.
Three-dimensional capturing of underwater archeological sites or sunken shipwrecks can support important documentation purposes. In this study, a novel 3D scanning system based on structured illumination is introduced, which supports cultural heritage documentation and measurement tasks in underwater environments. The newly developed system consists of two monochrome measurement cameras, a projection unit that produces aperiodic sinusoidal fringe patterns, two flashlights, a color camera, an inertial measurement unit (IMU), and an electronic control box. The opportunities and limitations of the measurement principles of the 3D scanning system are discussed and compared to other 3D recording methods such as laser scanning, ultrasound, and photogrammetry, in the context of underwater applications. Some possible operational scenarios concerning cultural heritage documentation are introduced and discussed. A report on application activities in water basins and offshore environments including measurement examples and results of the accuracy measurements is given. The study shows that the new 3D scanning system can be used for both the topographic documentation of underwater sites and to generate detailed true-scale 3D models including the texture and color information of objects that must remain under water.