Fraunhofer Institute for Computer Graphics Research

The current study evaluated the use of virtual reality (VR) and augmented reality (AR) platforms, developed within the scope of the SKILLS Integrated Project, for industrial maintenance and assembly (IMA) tasks training. VR and AR systems are now widely regarded as promising training platforms for complex and highly demanding IMA tasks. However, there is a need to empirically evaluate their efficiency and effectiveness compared to traditional training methods. Forty expert technicians were randomly assigned to four training groups in an electronic actuator assembly task: VR (training with the VR platform twice), Control-VR (watching a filmed demonstration twice), AR (training with the AR platform once), and Control-AR (training with the real actuator and the aid of a filmed demonstration once). A post-training test evaluated performance in the real task. Results demonstrate that, in general, the VR and AR training groups required longer training time compared to the Control-VR and Control-AR groups, respectively. There were fewer unsolved errors in the AR group compared to the Control-AR group, and no significant differences in final performance between the VR and Control-VR groups, probably due to a ceiling effect created by the use of two training trials in the selected task for participants who were expert technicians. The results suggest that use of the AR platform for training IMA tasks should be encouraged and use of the VR platform for that purpose should be further evaluated.

Abstract The analysis of large graphs plays a prominent role in various fields of research and is relevant in many important application areas. Effective visual analysis of graphs requires appropriate visual presentations in combination with respective user interaction facilities and algorithmic graph analysis methods. How to design appropriate graph analysis systems depends on many factors, including the type of graph describing the data, the analytical task at hand and the applicability of graph analysis methods. The most recent surveys of graph visualization and navigation techniques cover techniques that had been introduced until 2000 or concentrate only on graph layouts published until 2002. Recently, new techniques have been developed covering a broader range of graph types, such as time‐varying graphs. Also, in accordance with ever growing amounts of graph‐structured data becoming available, the inclusion of algorithmic graph analysis and interaction techniques becomes increasingly important. In this State‐of‐the‐Art Report, we survey available techniques for the visual analysis of large graphs. Our review first considers graph visualization techniques according to the type of graphs supported. The visualization techniques form the basis for the presentation of interaction approaches suitable for visual graph exploration. As an important component of visual graph analysis, we discuss various graph algorithmic aspects useful for the different stages of the visual graph analysis process. We also present main open research challenges in this field.

Abstract Interactive environments for dynamically deforming objects play an important role in surgery simulation and entertainment technology. These environments require fast deformable models and very efficient collision handling techniques. While collision detection for rigid bodies is well investigated, collision detection for deformable objects introduces additional challenging problems. This paper focuses on these aspects and summarizes recent research in the area of deformable collision detection. Various approaches based on bounding volume hierarchies, distance fields and spatial partitioning are discussed. In addition, image‐space techniques and stochastic methods are considered. Applications in cloth modeling and surgical simulation are presented.

The paper discusses Archeoguide which offers personalized augmented reality tours of archaeological sites. It uses outdoor tracking, mobile computing, 3D visualization and augmented reality techniques to enhance information presentation, reconstruct ruined sites, and simulate ancient life.

This article addresses the fundamentals of geometry-based watermarking. It presents a watermarking algorithm that modifies normal distribution to invisibly store information in the model's geometry.

Learning discriminative face features plays a major role in building high-performing face recognition models. The recent state-of-the-art face recognition solutions proposed to incorporate a fixed penalty margin on commonly used classification loss function, softmax loss, in the normalized hypersphere to increase the discriminative power of face recognition models, by minimizing the intra-class variation and maximizing the inter-class variation. Marginal penalty softmax losses, such as ArcFace and CosFace, assume that the geodesic distance between and within the different identities can be equally learned using a fixed penalty margin. However, such a learning objective is not realistic for real data with inconsistent inter-and intra-class variation, which might limit the discriminative and generalizability of the face recognition model. In this paper, we relax the fixed penalty margin constrain by proposing elastic penalty margin loss (ElasticFace) that allows flexibility in the push for class separability. The main idea is to utilize random margin values drawn from a normal distribution in each training iteration. This aims at giving the decision boundary chances to extract and retract to allow space for flexible class separability learning. We demonstrate the superiority of our ElasticFace loss over ArcFace and CosFace losses, using the same geometric transformation, on a large set of mainstream benchmarks. From a wider perspective, our ElasticFace has advanced the state-of-the-art face recognition performance on seven out of nine mainstream benchmarks. All training codes, pre-trained models, training logs will be publicly released1.

S.127-135

This paper presents the ARCHEOGUIDE project (Augmented Reality-based Cultural Heritage On-site GUIDE). ARCHEOGUIDE is an IST project, funded by the EU, aiming at providing a personalized electronic guide and tour assistant to cultural site visitors. The system provides on-site help and Augmented Reality reconstructions of ancient ruins, based on user's position and orientation in the cultural site, and realtime image rendering. It incorporates a multimedia database of cultural material for on-line access to cultural data, virtual visits, and restoration information. It uses multi-modal user interfaces and personalizes the flow of information to its user's profile in order to cater for both professional and recreational users, and for applications ranging from archaeological research, to education, multimedia publishing, and cultural tourism. This paper presents the ARCHEOGUIDE system and the experiences gained from the evaluation of an initial prototype by representative user groups at the archeological site of Olympia, Greece.

Visual analytics (VA) is typically applied in scenarios where complex data has to be analyzed. Unfortunately, there is a natural correlation between the complexity of the data and the complexity of the tools to study them. An adverse effect of complicated tools is that analytical goals are more difficult to reach. Therefore, it makes sense to consider methods that guide or assist users in the visual analysis process. Several such methods already exist in the literature, yet we are lacking a general model that facilitates in-depth reasoning about guidance. We establish such a model by extending van Wijk's model of visualization with the fundamental components of guidance. Guidance is defined as a process that gradually narrows the gap that hinders effective continuation of the data analysis. We describe diverse inputs based on which guidance can be generated and discuss different degrees of guidance and means to incorporate guidance into VA tools. We use existing guidance approaches from the literature to illustrate the various aspects of our model. As a conclusion, we identify research challenges and suggest directions for future studies. With our work we take a necessary step to pave the way to a systematic development of guidance techniques that effectively support users in the context of VA.

This paper considers the desired properties and possible applications of audio watermarking algorithms. Special attention is given to statistical methods working in the Fourier domain. It presents a solution to robust watermarking of audio data and reflects the security properties of the technique. Experimental results show good robustness of the approach against MP3 compression and other common signal processing manipulations. Enhancements to the presented methods are discussed.

Face image quality is an important factor to enable high-performance face recognition systems. Face quality assessment aims at estimating the suitability of a face image for the purpose of recognition. Previous work proposed supervised solutions that require artificially or human labelled quality values. However, both labelling mechanisms are error prone as they do not rely on a clear definition of quality and may not know the best characteristics for the utilized face recognition system. Avoiding the use of inaccurate quality labels, we proposed a novel concept to measure face quality based on an arbitrary face recognition model. By determining the embedding variations generated from random subnetworks of a face model, the robustness of a sample representation and thus, its quality is estimated. The experiments are conducted in a cross-database evaluation setting on three publicly available databases. We compare our proposed solution on two face embeddings against six state-of-the-art approaches from academia and industry. The results show that our unsupervised solution outperforms all other approaches in the majority of the investigated scenarios. In contrast to previous works, the proposed solution shows a stable performance over all scenarios. Utilizing the deployed face recognition model for our face quality assessment methodology avoids the training phase completely and further outperforms all baseline approaches by a large margin. Our solution can be easily integrated into current face recognition systems, and can be modified to other tasks beyond face recognition.

Big Data Analytics is an emerging field since massive storage and computing capabilities have been made available by advanced e-infrastructures. Earth and Environmental sciences are likely to benefit from Big Data Analytics techniques supporting the processing of the large number of Earth Observation datasets currently acquired and generated through observations and simulations. However, Earth Science data and applications present specificities in terms of relevance of the geospatial information, wide heterogeneity of data models and formats, and complexity of processing. Therefore, Big Earth Data Analytics requires specifically tailored techniques and tools. The EarthServer Big Earth Data Analytics engine offers a solution for coverage-type datasets, built around a high performance array database technology, and the adoption and enhancement of standards for service interaction (OGC WCS and WCPS). The EarthServer solution, led by the collection of requirements from scientific communities and international initiatives, provides a holistic approach that ranges from query languages and scalability up to mobile access and visualization. The result is demonstrated and validated through the development of lighthouse applications in the Marine, Geology, Atmospheric, Planetary and Cryospheric science domains.

Systems incorporating biometric technologies have become ubiquitous in personal, commercial, and governmental identity management applications. Both cooperative (e.g., access control) and noncooperative (e.g., surveillance and forensics) systems have benefited from biometrics. Such systems rely on the uniqueness of certain biological or behavioral characteristics of human beings, which enable for individuals to be reliably recognized using automated algorithms. Recently, however, there has been a wave of public and academic concerns regarding the existence of systemic bias in automated decision systems (including biometrics). Most prominently, face recognition algorithms have often been labeled as “racist” or “biased” by the media, nongovernmental organizations, and researchers alike. The main contributions of this article are: 1) an overview of the topic of algorithmic bias in the context of biometrics; 2) a comprehensive survey of the existing literature on biometric bias estimation and mitigation; 3) a discussion of the pertinent technical and social matters; and 4) an outline of the remaining challenges and future work items, both from technological and social points of view.

Biometric recognition technology has made significant advances over the last decade and is now used across a number of services and applications. However, this widespread deployment has also resulted in privacy concerns and evolving societal expectations about the appropriate use of the technology. For example, the ability to automatically extract age, gender, race, and health cues from biometric data has heightened concerns about privacy leakage. Face recognition technology, in particular, has been in the spotlight, and is now seen by many as posing a considerable risk to personal privacy. In response to these and similar concerns, researchers have intensified efforts towards developing techniques and computational models capable of ensuring privacy to individuals, while still facilitating the utility of face recognition technology in several application scenarios. These efforts have resulted in a multitude of privacy-enhancing techniques that aim at addressing privacy risks originating from biometric systems and providing technological solutions for legislative requirements set forth in privacy laws and regulations, such as GDPR. The goal of this overview paper is to provide a comprehensive introduction into privacy-related research in the area of biometrics and review existing work on Biometric Privacy-Enhancing Techniques (B-PETs) applied to face biometrics. To make this work useful for as wide of an audience as possible, several key topics are covered as well, including evaluation strategies used with B-PETs, existing datasets, relevant standards, and regulations and critical open issues that will have to be addressed in the future.

In recent years research on human activity recognition using wearable sensors has enabled to achieve impressive results on real-world data. However, the most successful activity recognition algorithms require substantial amounts of labeled training data. The generation of this data is not only tedious and error prone but also limits the applicability and scalability of today's approaches. This paper explores and systematically analyzes two different techniques to significantly reduce the required amount of labeled training data. The first technique is based on semi-supervised learning and uses self-training and co-training. The second technique is inspired by active learning. In this approach the system actively asks which data the user should label. With both techniques, the required amount of training data can be reduced significantly while obtaining similar and sometimes even better performance than standard supervised techniques. The experiments are conducted using one of the largest and richest currently available datasets.

In this paper, we evaluate several means of presenting route instructions to a mobile user. Starting from an abstract language-independent description of a route segment, we show how to generate various presentations for a mobile device ranging from spoken instructions to 3D visualizations. We then examine the relationship between the quality of positional information, available resources and the different types of presentations. The paper concludes with guidelines that help to determine which presentation to choose for a given situation

PURPOSE: We developed an effective way to precisely diagnose prostate cancer using a novel prostate biopsy system that integrates pre-interventional magnetic resonance imaging with peri-interventional ultrasound for perineal navigated prostate biopsy. MATERIALS AND METHODS: A total of 106 men with findings suspicious for prostate cancer (median age 66 years, prostate specific antigen 8.0 ng/ml and prostate volume 47 ml) underwent multiparametric 3 Tesla magnetic resonance imaging. Suspicious lesions were marked and data were transferred to the novel biopsy system. Using a custom-made biplane transrectal ultrasound probe mounted on a stepper we gathered 3-dimensional ultrasound data and fused them with magnetic resonance imaging data. As a result, suspicious magnetic resonance imaging lesions were superimposed over the transrectal ultrasound data. Three-dimensional biopsy planning was done, including systematic biopsies. Perineal biopsies were taken under live ultrasound guidance and the precise site of each biopsy was documented in 3 dimensions. We evaluated feasibility, safety and cancer detection. RESULTS: Prostate cancer was detected in 63 of 106 patients (59.4%). Magnetic resonance imaging findings correlated positively with histopathology in 71 of 103 patients (68.9%). In magnetic resonance imaging lesions marked as highly suspicious, the detection rate was 95.8% (23 of 24 cases). Lesion targeted cores had a significantly higher positivity rate than nontargeted cores. The procedural targeting error of the first 2,461 biopsy cores was 1.7 mm. Regarding adverse effects, 2 patients experienced urinary retention and 1 had a perineal hematoma. Urinary tract infections did not develop. CONCLUSIONS: Perineal stereotactic prostate biopsies guided by the combination of magnetic resonance imaging and ultrasound enable effective examination of suspicious magnetic resonance imaging lesions. Each biopsy core taken is documented accurately for its location in 3 dimensions, enabling magnetic resonance imaging validation and tailored treatment planning. The morbidity of the procedure was minimal.

Simulation-based training using VR techniques is a promising alternative to traditional training in minimally invasive surgery (MIS). Simulators let the trainee touch, feel, and manipulate virtual tissues and organs through the same surgical tool handles used in actual MIS while viewing images of tool-tissue interactions on a monitor as in real laparoscopic procedures.

The performance of face analysis and recognition systems depends on the quality of the acquired face data, which is influenced by numerous factors. Automatically assessing the quality of face data in terms of biometric utility can thus be useful to detect low-quality data and make decisions accordingly. This survey provides an overview of the face image quality assessment literature, which predominantly focuses on visible wavelength face image input. A trend towards deep learning-based methods is observed, including notable conceptual differences among the recent approaches, such as the integration of quality assessment into face recognition models. Besides image selection, face image quality assessment can also be used in a variety of other application scenarios, which are discussed herein. Open issues and challenges are pointed out, i.a., highlighting the importance of comparability for algorithm evaluations and the challenge for future work to create deep learning approaches that are interpretable in addition to providing accurate utility predictions.

The manual assessment of Activities of Daily Living (ADLs) is a fundamental problem in elderly care. The use of miniature sensors placed in the environment or worn by a person has great potential in effective and unobtrusive long term monitoring and recognition of ADLs. This paper presents an effective and unobtrusive activity recognition system based on the combination of the data from two different types of sensors: RFID tag readers and accelerometers. We evaluate our algorithms on non-scripted datasets of 10 housekeeping activities performed by 12 subjects. The experimental results show that recognition accuracy can be significantly improved by fusing the two different types of sensors. We analyze different acceleration features and algorithms, and based on tag detections we suggest the best tags’ placements and the key objects to be tagged for each activity.