The use of rescue robots in endangered buildings avoids risking lives. A scene representation for orientation and navigation purposes that is visually understandable and reduces the storage space is needed. Researchers have proposed recovering primitive shapes from different kinds of 3D data. However, most of them focus on recovering individual objects, not complete scenes. Therefore, we propose a combination of planes, superquadrics, and points to abstract scenes from depth maps.The introduced approach consists of the following steps. First, prominent planes in the input data are detected, and their corresponding points are removed. Second, the remaining points are segmented, and a superquadric or a set of superquadrics is recovered from each segment. Finally, the remaining points, the detected planes, and the recovered superquadrics are used to create a parsimonious representation of the input.This approach also introduces a new objective function to recover superquadrics, which provides better results for this application than the existing objective functions. Additionally, a constraint for recovering superquadrics from depth images is proposed, ensuring results consistent with reality.

extended-abstract Share on Gone But Not Forgotten: Evaluating Performance and Scalability of Real-Time Mesoscopic Agents Authors: Sven Seele Department of Computer Science and Institute of Visual Computing, University of Applied Sciences, Bonn-Rhein-Sieg Sankt Augustin, Germany Department of Computer Science and Institute of Visual Computing, University of Applied Sciences, Bonn-Rhein-Sieg Sankt Augustin, GermanyView Profile , Luisa Pätzold Department of Computer Science and Institute of Visual Computing, University of Applied Sciences, Bonn-Rhein-Sieg Sankt Augustin, Germany Department of Computer Science and Institute of Visual Computing, University of Applied Sciences, Bonn-Rhein-Sieg Sankt Augustin, GermanyView Profile , Rainer Herpers Department of Computer Science and Institute of Visual Computing, University of Applied Sciences, Bonn-Rhein-Sieg Sankt Augustin, Germany Department of Computer Science and Institute of Visual Computing, University of Applied Sciences, Bonn-Rhein-Sieg Sankt Augustin, GermanyView Profile Authors Info & Claims IVA '20: Proceedings of the 20th ACM International Conference on Intelligent Virtual AgentsOctober 2020 Article No.: 50Pages 1–3https://doi.org/10.1145/3383652.3423891Published:19 October 2020Publication History 0citation55DownloadsMetricsTotal Citations0Total Downloads55Last 12 Months10Last 6 weeks1 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

In this work, we automatically distinguish the efficient high elbow pose from dropping one in pulling phase of front crawl stroke in front view amateurly recorded videos. This task is challenging due to the aquatic environment and missing depth information. We predict the pull's efficiency through multiclass svm and random forest classifiers given arms key positions and angles as the feature set. We evaluate our approach over a labeled dataset of video frames taken from 25 members of masters' swim club at Ryerson University with different levels of expertise and physiological characteristics. Our results show the effectiveness of our approach with random forest classifier, yielding 67% accuracy.

We present a multi-user virtual reality (VR) setup that aims at providing novel training tools for paramedics that enhances current learning methods. The hardware setup consists of a two-user full-scale VR environment with head-mounted displays for two interactive trainees and one additional desktop pc for one trainer participant. The software provides a connected multi-user environment, showcasing a paramedic emergency simulation with focus on anaphylactic shock, a representative scenario for critical medical cases that happen too rare to eventually occur within a regular curricular term of vocational training. The prototype offers hands-on experience on multi-user VR in an applied scenario, generating discussion around current state and future development concerning three important research areas: (a) user navigation, (b) interaction, (c) level of visual abstraction, and (d) level of task abstraction.

Underlying semantics are important to facilitate traffic simulations in virtual environments. The work presented here introduces an universal and extensible model for the representation of road network logics. Additionally, setup processes of road network logics following the model are described, and the integration of different traffic simulation approaches is discussed. For evaluation, scenes according to different scenarios were created and semantics were integrated. Results indicate significant time savings from automatic generation of road network logics.

Developing applications for Field Programmable Gate Array (FPGA) devices utilizes Computer Aided Design (CAD) flows. The transition from a high level Verilog hardware description to the optimized structure of programmed soft logic blocks and routing structure includes stages such as Verilog synthesis, hardware mapping, logical synthesis, packing, placement and routing. The VTR CAD flow is a collaborative project consisting of Odin II (University of New Brunswick), ABC (University of California, Berkeley) and VPR (University of Toronto), which offers an FPGA CAD flow for research and experimentation purposes. This paper describes developments in the visualization and simulation modules of Odin II, the first stage of the CAD flow. The contributions include new netlist visualization possibilities as well as an extended netlist simulator capable of simulating circuits with multiple clocks and providing extended generic structure simulation abilities. This results in the possibility to explore and simulate a larger set of new FPGA architectures and evaluate them using the VTR flow.

In interactive visualization environments which use a multiple screen setup, every output device has to be supplied frequently with video information. Such virtual environments often use large projection screens, which require high resolution video data. When increasing scene complexity, it can become a challenge to equip a single computer with graphics hardware powerful enough for the task. An efficient approach is to distribute the workload to multiple low-cost computer systems such as game consoles. Nowadays' game consoles are very powerful and specialized for interactive graphics applications, therefore they are suitable to being used for rendering purposes. A framework has been developed that builds on Microsoft's XNA Game Studio. It enables for interactive distributed rendering on multiple Xbox 360 systems and PCs. Tasks such as game logic synchronization, network session management are fully handled by the framework. A game built with it can focus mostly on its own implementation. The framework's structure follows that of the XNA Game Studio, which allows porting existing game projects quickly. Our evaluation showed that the framework is a lightweight solution which leaves almost the full CPU time to the actual game.

This paper presents the implementation and validation of a tracking approach for image processing in hardware. It compares the implementation for the addressed problem on a Field Programmable Gate Array (FPGA) with a software implementation for a General Purpose Processor (GPP) architecture. For both solutions the implementation costs for their development is an important aspect in the validation. This research project is motivated by the MI6 project of the Computer Vision research group, which is located at the Bonn-Rhein-Sieg University of Applied Sciences. The intent of the MI6 project is the tracking of a user in an immersive environment. In this research work the development and validation of a detection system for BLOBs on a Cyclone II FPGA from Altera has been implemented. The analysis of the hardware solution showed a similar precision for the tracking as the software approach. One problem is the large increase of allocated resources when extending the system to process more objects. The implementation of the tracking approach in hardware required much more effort than the software solution. The design of high level problems in hardware for this case are more expensive than the software implementation. The results of the implementation indicate that a mixed hardware/software solution would provide optimal results.

In this paper we present the design, implementation, and testing of an evaluation tool for the ongoing development of the Prosthetic Device Communication Protocol (PDCP) which is an open protocol and is featured in the University of New Brunswick's most recent prosthetic limb research project, the UNB Hand System. This prosthetic device utilizes the CAN bus hardware with the PDCP for passing command and data messages between modules within the prosthetic limb system. The PDCP allows abstraction of the underlying bus system and allows different network topologies depending on particular needs. To be able to analyze communication in the CAN layers as well as in the PDCP layer we present our own solutions utilizing an FPGA for CAN bus bandwidth load monitoring and a microcontroller for PDCP monitoring and analysis.

Field Programmable Gate Arrays (FPGA) are used in many fields of research, e.g. to create prototypes of hardware or in applications where hardware functionality has to be changed more frequently. Boolean circuits, which can be implemented by FPGAs are the compiled result of hardware description languages such as Verilog or VHDL. Odin II is a tool, which supports developers in the research of FPGA based applications and FPGA architecture exploration by providing a framework for compilation and verification. In combination with the tools ABC, T-VPACK and VPR, Odin II is part of a CAD flow, which compiles Verilog source code that targets specific hardware resources. This paper describes the development of a graphical user interface as part of Odin II. The goal is to visualize the results of these tools in order to explore the changing structure during the compilation and optimization processes, which can be helpful to research new FPGA architectures and improve the work flow.

In interactive graphics it is often necessary to introduce large changes in the image in response to updated information about the state of the system. Updating the local state immediately would lead to a sudden transient change in the image, which could be perceptually disruptive. However, introducing the correction gradually using smoothing operations increases latency and degrades precision. It would be beneficial to be able to introduce graphic updates immediately if they were not perceptible. In the paper the use of saccade-contingent updates is exploited to hide graphic updates during the period of visual suppression that accompanies a rapid, or saccadic, eye movement. Sensitivity to many visual stimuli is known to be reduced during a change in fixation compared to when the eye is still. For example, motion of a small object is harder to detect during a rapid eye movement (saccade) than during a fixation. To evaluate if these findings generalize to large scene changes in a virtual environment, gaze behavior in a 180 degree hemispherical display was recorded and analyzed. This data was used to develop a saccade detection algorithm adapted to virtual environments. The detectability of trans-saccadic scene changes was evaluated using images of high resolution real world scenes. The images were translated by 0.4, 0.8 or 1.2 degrees of visual angle during horizontal saccades. The scene updates were rarely noticeable for saccades with a duration greater than 58 ms. The detection rate for the smallest translation was just 6.25%. Qualitatively, even when trans-saccadic scene changes were detectible, they were much less disturbing than equivalent changes in the absence of a saccade.

The application of an elastic graph-matching approach to discriminate facial image regions is presented. In contrast to the dynamic link architecture introduced by the Malsburg group, our application is not an identification task but a classification task. Therefore, our approach differs in several important aspects: (1) the choice of the filter set, (2) the selection of the positions of the nodes of the graph to represent the characteristic image information, (3) the generation of a representative reference pattern needed for the calculation of the classifications, and (4) a new two-step graph-matching approach based on the simulated annealing technique. The approach was tested on facial regions taking the eye region as an example target. A classification performance for the verification of eye regions of more than 93% was achieved.

Usually, the first processing step in computer vision systems consists of a spatial convolution with only a few simple filters. Therefore, information is lost of it is not represented explicitly for the following processing steps. This paper proposes a new hierarchical filter scheme that can efficiently synthesize the responses for a large number of specific filters. The scheme is based on steerable filters. It also allows for an efficient on-line adjustment of the trade off between the speed and the accuracy of the filters. We apply this method to the detection of facial keypoints, especially the eye corners. These anatomically defined keypoints exhibit a large variability in their corresponding image structures so that a flexible low level feature extraction is required.

In this contribution Dynamic Cell Structures (DCS network) are applied to classify local image structures at particular facial landmarks. The facial landmarks such as the corners of the eyes or intersections of the iris with the eyelid are computed in advance by a combined model and data driven sequential search strategy. To reduce the detection error after the processing of the sequential search strategy, the computed image positions are verified applying a DCS network. The DCS network is trained by supervised learning with feature vectors which encode spatially arranged edge and structural information at the keypoint position considered. The model driven localization as well as the data driven verification are based on steerable filters, which build a representation comparable with one provided by a receptive field in the human visual system. We apply a DCS based classifier because of its ability to grasp the topological structure of complex input spaces and because it has proved successful in a number of other classification tasks. In our experiments the average error resulting from false positive classifications is less than 1%.

The authors introduce a method for the automatic detection of facial features and characteristic anatomical keypoints. In the application they are aiming at the anatomical landmarks are used to accurately measure facial features. Their approach is essentially bused on a selective search and sequential tracking of characteristic edge and line structures of the facial object to be searched. It integrates model knowledge to guarantee a consistent interpretation of the abundance of local features. The search and the tracking is controlled in each step by interpreting the already derived edge and line information in the context of the whole considered region. For their application, the edge and line detection has to be very precise and flexible. Therefore, they apply a powerful filtering scheme based on steerable filters.

Facial keypoints such as eye corners are important features for a number of different tasks in automatic face processing. The problem is that facial keypoints rather have an anatomical high-level definition than a low-level one. Therefore, they cannot be detected reliably by purely data-driven methods like corner detectors that are only based on the image data of the local neighborhood. In this contribution we introduce a method for the automatic detection of facial keypoints. The method integrates model knowledge to guarantee a consistent interpretation of the abundance of local features. The detection is based on a selective search and sequential tracking of edges controlled by model knowledge. For this, the edge detection has to be very flexible. Therefore, we apply a powerful filtering scheme based on steerable filters.