3D Action Research Articles

Simultaneous tracking and action recognition for single actor human actions

This paper presents an approach to simultaneously tracking the pose and recognizing human actions in a video. This is achieved by combining a Dynamic Bayesian Action Network (DBAN) with 2D body part models. Existing DBAN implementation relies on fairly weak observation features, which affects the recognition accuracy. In this work, we use a 2D body part model for accurate pose alignment, which in turn improves both pose estimate and action recognition accuracy. To compensate for the additional time required for alignment, we use an action entropy-based scheme to determine the minimum number of states to be maintained in each frame while avoiding sample impoverishment. In addition, we also present an approach to automation of the keypose selection task for learning 3D action models from a few annotations. We demonstrate our approach on a hand gesture dataset with 500 action sequences, and we show that compared to DBAN our algorithm achieves 6% improvement in accuracy.

Relation between the 4d superconformal index and the S 3 partition function

A relation between the 4d superconformal index and the S^3 partition function is studied with focus on the 4d and 3d actions used in localization. In the case of vanishing Chern-Simons levels and round S^3 we explicitly show that the 3d action is obtained from the 4d action by dimensional reduction up to terms which do not affect the exact results. By combining this fact and a recent proposal concerning a squashing of S^3 and SU(2) Wilson line, we obtain a formula which gives the partition function depending on the Weyl weight of chiral multiplets, real mass parameters, FI parameters, and a squashing parameter as a limit of the index of a parent 4d theory.

Elasto-Plastic Analysis of Seismic Response of Valve-Hall Structure with Suspension Valves of Large-Scale Converter Station

Elasto-plastic analysis of seismic responses of valve hall structures were carried out by using finite element software, and the effect of seismic waves on the seismic responses of the valve hall structures and suspension equipments were studied. Results show that significant torsional responses of the structure can be found under longitudinal and 3D earthquake actions. Under 3D earthquake actions, the seismic responses of the suspension valves are much more significant than those under 1D earthquake actions, the maximum tensile force of the suspenders is about twice of that under 1D action. The seismic responses of the suspension valves under vertical earthquake actions are much stronger than those under horizontal earthquake actions, when suffering strong earthquake actions; the maximum vertical acceleration of the suspension valves is about 4 times of that under horizontal earthquake actions. It is recommended that the effects of 3D earthquake actions on the structure should be considered in seismic response analysis of the valve hall structure.

Three‐Dimensional Entertainment as a Novel Cause of Takotsubo Cardiomyopathy

Takotsubo cardiomyopathy (TC) is an uncommon entity. It is known to occur in the setting of extreme catecholamine release and results in left ventricular dysfunction without evidence of angiographically definable coronary artery disease. There have been no published reports of TC occurring with visual stimuli, specifically 3-dimensional (3D) entertainment. We present a 55-year-old woman who presented to her primary care physician's office with extreme palpitations, nausea, vomiting, and malaise <48 hours after watching a 3D action movie at her local theater. Her electrocardiogram demonstrated ST elevations in aVL and V1, prolonged QTc interval, and T-wave inversions in leads I, II, aVL, and V2-V6. Coronary angiography revealed angiographically normal vessels, elevated left ventricular filling pressures, and decreased ejection fraction with a pattern of apical ballooning. The presumed final diagnosis was TC, likely due to visual-auditory-triggered catecholamine release causing impaired coronary microcirculation.

Wireless cardiac action potential transmission with ultrasonically inserted silicon microprobes

This paper reports on the integration of ultrasonically inserted horn-shaped cardiac probes with wireless transmission of 3D cardiac action potential measurement for applications in ex vivo preparations such as monitoring the onset of ventricular fibrillation. Ultrasonically inserted silicon horn probes permit reduced penetration force during insertion, allowing silicon, a brittle material, to penetrate cardiac tissue. The probes also allow recording from multiple sites that are lithographically defined. An application-specific integrated circuit has been designed with a 40 dB amplifying stage and a frequency modulating oscillator at 95 MHz to wirelessly transmit the recorded action potentials. This ultrasonically inserted microprobe wireless system demonstrates the initial results in wireless monitoring of 3D action potential propagation, and the extraction of parameters of interest including the action potential duration and diastolic interval.

Beyond the classic correction system: a numerical nonrigid approach to the scoliosis brace

Background context Adolescent idiopathic scoliosis (AIS) causes a spine and rib cage three-dimensional (3D) deformity previously treated by bracing. Whatever the manufacturing process, this rigid system acts biomechanically on the patient through the “three-point bending” mechanical principle. It applies corrective forces to a limited area and acts especially in the frontal plane. It seemed to us that a nonrigid system, called “Cbrace,” with 3D action allowing distribution of forces could increase compliance and provide better long-term correction prospects. Purpose The aim of this study was to design a nonrigid brace by numerically testing in a finite-element model developed here. Study design A finite-element model has been developed to simulate brace effect on AIS right thoracic deformation of a 10-year-old patient. Methods A two-step method was needed; first, the reliability of our model is evaluated, and then, the ability to use distributed forces to correct scoliosis deformation is tested. To obtain a 3D correction, several treatments are experimented, leading to a comparison test between the best combination to the “three-point bending” principle. Results The numerical model developed here shows good qualitative answers for the treatment of brace forces. The first results demonstrate numerically that distributed forces may be of interest in brace treatment design. Overall force of 40 N above cartilage of the last nonfloating ribs associated to two posterior asymmetrical areas appears to be the best way to correct scoliosis deformation with nonrigid action. Conclusion The results show numerical efficacy of distributed forces to correct spinal deformities and raises the prospect that a new numerical brace, called “Cbrace,” could be a starting point in the search for a nonrigid system.

An empirical pipeline to derive gaze prediction heuristics for 3D action games

Gaze analysis and prediction in interactive virtual environments, such as games, is a challenging topic since the 3D perspective and variations of the viewpoint as well as the current task introduce many variables that affect the distribution of gaze. In this article, we present a novel pipeline to study eye-tracking data acquired from interactive 3D applications. The result of the pipeline is an importance map which scores the amount of gaze spent on each object. This importance map is then used as a heuristic to predict a user's visual attention according to the object properties present at runtime. The novelty of this approach is that the analysis is performed in object space and the importance map is defined in the feature space of high-level properties. High-level properties are used to encode task relevance and other attributes, such as eccentricity, which may have an impact on gaze behavior. The pipeline has been tested with an exemplary study on a first-person shooter game. In particular, a protocol is presented describing the data acquisition procedure, the learning of different importance maps from the data, and finally an evaluation of the performance of the derived gaze predictors. A metric measuring the degree of correlation between attention predicted by the importance map and the actual gaze yielded clearly positive results. The correlation becomes particularly strong when the player is attentive to an in-game task.

A framework for quantifying properties of 3-dimensional movement-based signals

Abstract Understanding how signal properties are optimized for the reliable transmission of information requires accurate description of the signal in time and space. For movement-based signals where movement is restricted to a single plane, measurements from a single viewpoint can be used to consider a range of viewing positions based on simple geometric calculations. However, considerations of signal properties from a range of viewing positions for movements extending into three-dimensions (3D) are more problematic. We present here a new framework that overcomes this limitation, and enables us to quantify the extent to which movement-based signals are view-specific. To illustrate its application, a Jacky lizard tail flick signal was filmed with synchronized cameras and the position of the tail tip digitized for both recordings. Camera alignment enabled the construction of a 3D display action pattern profile. We analyzed the profile directly and used it to create a detailed 3D animation. In the virtual environment, we were able to film the same signal from multiple viewing positions and using a computational motion analysis algorithm (gradient detector model) to measure local image velocity in order to predict view dependent differences in signal properties. This approach will enable consideration of a range of questions concerning movement-based signal design and evolution that were previously out of reach.

Neural basis for associative face memory in the anterior inferior temporal cortex of macaque monkeys

In this paper we focus on the joint problem of tracking humans and recognizing human action in scenarios such as a kitchen scenario or a scenario where a robot cooperates with a human, e.g., for a manufacturing task. In these scenarios, the human directly interacts with objects physically by using/manipulating them or by, e.g., pointing at them such as in “Give me that…”. To recognize these types of human actions is difficult because (a) they ought to be recognized independent of scene parameters such as viewing direction and (b) the actions are parametric, where the parameters are either object-dependent or as, e.g., in the case of a pointing direction convey important information. One common way to achieve recognition is by using 3D human body tracking followed by action recognition based on the captured tracking data. For the kind of scenarios considered here we would like to argue that 3D body tracking and action recognition should be seen as an intertwined problem that is primed by the objects on which the actions are applied. In this paper, we are looking at human body tracking and action recognition from a object-driven perspective. Instead of the space of human body poses we consider the space of the object affordances, i.e., the space of possible actions that are applied on a given object. This way, 3D body tracking reduces to action tracking in the object (and context) primed parameter space of the object affordances. This reduces the high-dimensional joint-space to a low-dimensional action space. In our approach, we use parametric hidden Markov models to represent parametric movements; particle filtering is used to track in the space of action parameters. We demonstrate its effectiveness on synthetic and on real image sequences using human-upper body single arm actions that involve objects.

Learning Generative Models for Multi-Activity Body Pose Estimation

We present a method to simultaneously estimate 3D body pose and action categories from monocular video sequences. Our approach learns a generative model of the relationship of body pose and image appearance using a sparse kernel regressor. Body poses are modelled on a low-dimensional manifold obtained by Locally Linear Embedding dimensionality reduction. In addition, we learn a prior model of likely body poses and a dynamical model in this pose manifold. Sparse kernel regressors capture the nonlinearities of this mapping efficiently. Within a Recursive Bayesian Sampling framework, the potentially multimodal posterior probability distributions can then be inferred. An activity-switching mechanism based on learned transfer functions allows for inference of the performed activity class, along with the estimation of body pose and 2D image location of the subject. Using a rough foreground segmentation, we compare Binary PCA and distance transforms to encode the appearance. As a postprocessing step, the globally optimal trajectory through the entire sequence is estimated, yielding a single pose estimate per frame that is consistent throughout the sequence. We evaluate the algorithm on challenging sequences with subjects that are alternating between running and walking movements. Our experiments show how the dynamical model helps to track through poorly segmented low-resolution image sequences where tracking otherwise fails, while at the same time reliably classifying the activity type.

Performance Evaluation of the Philips “Gemini” PET/CT System

Methods to test the practical performance of the Philips Gemini PET/CT system during clinical imaging are described and results presented. The test methods used were based on the NEMA standards and those suggested by a task group of the Commission of European Communities (EEC). These were performed using the EEC emission phantom, and phantoms provided by Philips. Scans were acquired using the routine clinical protocols. Images were reconstructed using a 3D row action maximum likelihood algorithm (RAMLA). Attenuation correction (AC) was applied using CT data (CTAC) and Cs-137 source transmission scans (CsAC). Images can be acquired with the PET and CT gantries closed or open. Results of tests on the gantry alignment show that the static offsets varied by a maximum of 4.5 mm when the system was first installed, but as the floor has settled these variations have decreased to a maximum of 2.5 mm. The mean negative nonuniformity was $-4.8 pm 1.1hbox%$ and $-5.3 pm 1.1hbox%$ for CsAC and CTAC respectively and the mean positive nonuniformity was $4.7 pm 1.4hbox%$ and $5.5 pm 1.6hbox%$ . Scatter fraction measurements were made in the EEC body phantom. The reconstructed scatter fractions were 0.18 and 0.16 with and without out-of-field-of-view activity respectively ( $pm$ OFOVA) for both CsAC and CTAC images. The sinogram-derived (NEMA NU 2-1994) scatter fraction was 0.44 and 0.43 $pm$ OFOVA. At the centre of the field of view (FOV) the transverse spatial resolution (SR) was 5.0 mm in air (5.0 mm in water) and the axial SR was 6.0 mm (6.8 mm). 10 cm away from the centre the transverse radial SR was 5.4 mm (5.3 mm), the transverse tangential SR was 5.2 mm (5.4 mm) and the axial SR was 6.6 mm (7.2 mm). The recovery coefficients demonstrate the partial volume effect. The AC accuracy was similar in water (both $0.2 pm 0.1 hbox%$ ) and Teflon [ $10.9 pm 1.9hbox%$ (CTAC) and $10.0 pm 2.0hbox%$ (CsAC)], for both types of AC, but very different in air [ $14.3 pm 2.2hbox%$ (CTAC) and $31.3pm 1.5hbox%$ (CsAC)]. If the transmission maps were not segmented prior to AC these gave much higher and very different residual AC errors. The peak noise equivalent count rate was 45.2 kcps at 184 MBq and 25.2 kcps at 127 MBq $pm$ OFOVA. The system sensitivity to true and scattered counts was 34.1 kcps/(kBq/cm $^3$ ) and to true counts was 19.4 kcps/(kBq/cm $^3$ ).

Development of an oscillating/rotating/pulsating toothbrush: The Oral-B professionalCare™ series

The Oral-B range of power toothbrushes are based on a clinically proven oscillating/rotating action. This review considers the extensive series of laboratory and clinical studies involved in the development of the Oral-B power toothbrushes with three-dimensional (3D) brush head action (i.e. oscillating/rotating/pulsating), now known as the Oral-B ProfessionalCaretrade mark Series. The unique, highly effective cleaning performance of the 3D action incorporated in the D15 and D17 models is proven in both the laboratory and clinical environments, thereby establishing an evidence-based platform for the development of the most recent ProfessionalCare 7000. In the early development phase, in-vitro studies indicated that the additional pulsating action and advancements in filament technology improved approximal penetration, where plaque removal is most difficult. In-vitro models provide a useful development tool for the rapid assessment of design modifications, while controlling many factors that confound in-vivo studies, but the relevance of their findings needs to be confirmed in the clinical situation by direct improvements in oral health. Results from well-controlled, randomised clinical studies in the development programme have demonstrated the superiority of the D15 and D17 models with 3D action over a manual toothbrush and other power toothbrushes of various actions and designs in the improvement of oral health outcomes such as plaque removal (especially from approximal sites), reduction of gingival inflammation, and control of calculus and stain formation, with no greater potential to cause oral tissue abrasion. Furthermore, user acceptance of the oscillating/rotating/pulsating toothbrush may encourage long-term compliance during normal use.

Three-Dimensional Effects on Openings of Laterally Loaded Pierced Shear Walls

Current design provisions comprise broadly described information for the detailing of reinforcement around the openings of pierced shear walls. To address this deficiency, the load capacity and stress distribution around the openings were analyzed by conducting three-dimensional (3D) nonlinear pushover analyses on typical shear wall dominant building structures. The diaphragm flexibility, behavior of transverse walls and slab-wall interaction during the 3D action were investigated in addition to effects of 3D and 2D modeling on the capacity evaluation. An effort was spent to illuminate the significance of different size and location of openings within the pierced walls having variable reinforcement ratios. The results of this study indicated that the stress flow and crack patterns around the openings of the 3D cases were drastically different than those computed for the 2D cases. The tension.compression coupling effects caused by the wall-to-wall and wall-to-slab interactions provided a significant contribution for increasing the global lateral resistance.

Geometry mediated supersymmetry breaking

We investigate SUSY breaking mediated through the deformation of the space-time geometry due to the backreaction of a nontrivial configuration of a bulk scalar field. To illustrate its features, we work with a toy model in which the bulk is four dimensions. Using the superconformal formulation of SUGRA, we provide a systematic method of deriving the 3D effective action expressed by the superfields, which can basically be extended to 5D SUGRA straightforwardly.

Spherical and planar three-dimensional anti-de Sitter black holes

The technique of dimensional reduction was used in a recent paper (Zanchin V T, Kleber A and Lemos J P S 2002 Phys. Rev. D 66 064022) where a three-dimensional (3D) Einstein–Maxwell–dilaton theory was built from the usual four-dimensional (4D) Einstein–Maxwell–Hilbert action for general relativity. Starting from a class of 4D toroidal black holes in asymptotically anti-de Sitter (AdS) spacetimes several 3D black holes were obtained and studied in such a context. In the present work we choose a particular case of the 3D action which presents Maxwell field, dilaton field and an extra scalar field, besides gravity field and a negative cosmological constant, and obtain new 3D static black hole solutions whose horizons may have spherical or planar topology. We show that there is a 3D static spherically symmetric solution analogous to the 4D Reissner–Nordström–AdS black hole, and obtain other new 3D black holes with planar topology. From the static spherical solutions, new rotating 3D black holes are also obtained and analysed in some detail.

CERN LEP2 constraint on 4D QED having a dynamically generated spatial dimension

We study 4D QED in which one spatial dimension is dynamically generated from 3D action, following the mechanism proposed by Arkani-Hamed, Cohen and Georgi. In this model, the generated fourth dimension is discretized by an interval parameter $a$. We examine the phenomenological constraint on the parameter $a$ coming from the collider experiments on the QED process $e^+ e^- \to \gamma \gamma$. It is found that the LEP2 experiments give the constraint of $1/a \simgt 461\gev$. Expected bound on the same parameter $a$ at the future $e^+ e^-$ linear collider is briefly discussed.

Nonlinear seismic response evaluation of tunnel form building structures

The occurrences of ( M w=7.4) Kocaeli and ( M w=7.1) Duzce earthquakes in Turkey in 1999 once again demonstrate the nondamaged and high performance conditions of RC shear wall dominant structures commonly built by using the tunnel form technique. This study presents their seismic performance evaluation based on the nonlinear pushover analyses of two case studies. The contribution of transverse walls and slab-wall interaction during the 3D action, the effects of 3D and 2D modeling on the capacity–demand relation, as well as diaphragm flexibility, torsion and damping effects were investigated. An effort was spent to develop a shell element having closing–opening and rotating crack capabilities. This study shows that the applied methodology has a considerable significance for predicting the actual capacity, failure mechanism, and evaluation of the seismic response of tunnel form buildings.

Application of the row action maximum likelihood algorithm with spherical basis functions to clinical PET imaging

Three-dimensional (3D) reconstructions from fully 3D positron emission tomography (PET) data can yield high-quality images but at a high computational cost. The 3D row action maximum likelihood algorithm (3D RAMLA) with spherically-symmetric basis functions (blobs) has recently been modified to reconstruct multi-slice 2D PET data after Fourier rebinning (FORE) but still using 3D basis functions (2.5D RAMLA. In this study 2.5D RAMLA and 3D RAMLA were applied to several patient and phantom PET datasets to assess their clinical performance. RAMLA performance was compared to that for the reconstruction techniques in routine clinical use on the authors' PET scanners. Torso phantom and whole-body patient scans acquired on the C-PET scanner were reconstructed after FORE with filtered back-projection (FORE+FBP), the ordered subsets expectation maximization algorithm (FORE+OSEM), and FORE+2.5D RAMLA for various reconstruction parameters. The 3D Hoffman brain phantom scanned on the HEAD Penn-PET scanner was reconstructed with the 3D reprojection algorithm (3DRP) and 3D RAMLA, as well, as FORE+FBP, FORE+OSEM, and FORE+2.5D RAMLA. The authors' results demonstrate improvement of 3D and 2.5D RAMLA with blob basis functions, compared to the reconstruction methods currently in clinical use, in terms of contrast recovery and noise, especially in regions of limited statistics.

Intermittency in the N-component φ4 theory

The effective 3D action of the N-component φ 4 theory is adapted to the space-time hyperbolas of the critical multiparticle system, produced in relativistic ion collisions. In a systematic treatment of the ordered phase at T = T c , a strong intermittency pattern, with a minimal scale in rapidity, naturally emerges. The connection of the solution to chiral QCD phase transition is emphasized and its phenomenological implications are discussed.

A self consistent study of the phase transition in the scalar electroweak theory at finite temperature

We propose the study of the phase transition in the scalar electroweak theory at finite temperature by a two - step method. It combines i) dimensional reduction to a 3-dimensional {\it lattice\/} theory via perturbative blockspin transformation, and ii) either further real space renormalization group transformations, or solution of gap equations, for the 3d lattice theory. A gap equation can be obtained by using the Peierls inequality to find the best quadratic approximation to the 3d action. % This method avoids the lack of self consistency of the usual treatments which do not separate infrared and UV-problems by introduction of a lattice cutoff. The effective 3d lattice action could also be used in computer simulations.