Dynamic Point Research Articles

Two-stream Multi-level Dynamic Point Transformer for Two-person Interaction Recognition

As a fundamental aspect of human life, two-person interactions contain meaningful information about people’s activities, relationships, and social settings. Human action recognition serves as the foundation for many smart applications, with a strong focus on personal privacy. However, recognizing two-person interactions poses more challenges due to increased body occlusion and overlap compared to single-person actions. In this paper, we propose a point cloud-based network named Two-stream Multi-level Dynamic Point Transformer for two-person interaction recognition. Our model addresses the challenge of recognizing two-person interactions by incorporating local-region spatial information, appearance information, and motion information. To achieve this, we introduce a designed frame selection method named Interval Frame Sampling (IFS), which efficiently samples frames from videos, capturing more discriminative information in a relatively short processing time. Subsequently, a frame features learning module and a two-stream multi-level feature aggregation module extract global and partial features from the sampled frames, effectively representing the local-region spatial information, appearance information, and motion information related to the interactions. Finally, we apply a transformer to perform self-attention on the learned features for the final classification. Extensive experiments are conducted on two large-scale datasets, the interaction subsets of NTU RGB+D 60 and NTU RGB+D 120. The results show that our network outperforms state-of-the-art approaches in most standard evaluation settings.

(3+1)-dimensional Pearcey-Gaussian wave packet with arbitrary velocity driven by flying focus.

The group velocity (GV) modulation of space-time wave packets (STWPs) along the transverse and longitudinal directions in free space is constrained by various factors. To surmount this limitation, a technique called "flying focus" has been developed, which enables the generation of laser pulses with dynamic focal points that can propagate at arbitrary velocities independent of GV. In this Letter, we propose a (3+1)-dimensional Pearcey-Gauss wave packet based on the "flying focus" technique, which exhibits superluminal propagation, transverse focus oscillation, and longitudinal periodic autofocusing. By selecting appropriate parameters, we can flexibly manipulate the position, the size, and the number of focal points- or make the wave packet follow a desired trajectory. This work may pave the way for the advancement of space-time structured light fields.

Dynamic 3D Point Cloud Sequences as 2D Videos.

Dynamic 3D point cloud sequences serve as one of the most common and practical representation modalities of dynamic real-world environments. However, their unstructured nature in both spatial and temporal domains poses significant challenges to effective and efficient processing. Existing deep point cloud sequence modeling approaches imitate the mature 2D video learning mechanisms by developing complex spatio-temporal point neighbor grouping and feature aggregation schemes, often resulting in methods lacking effectiveness, efficiency, and expressive power. In this paper, we propose a novel generic representation called Structured Point Cloud Videos (SPCVs). Intuitively, by leveraging the fact that 3D geometric shapes are essentially 2D manifolds, SPCV re-organizes a point cloud sequence as a 2D video with spatial smoothness and temporal consistency, where the pixel values correspond to the 3D coordinates of points. The structured nature of our SPCV representation allows for the seamless adaptation of well-established 2D image/video techniques, enabling efficient and effective processing and analysis of 3D point cloud sequences. To achieve such re-organization, we design a self-supervised learning pipeline that is geometrically regularized and driven by self-reconstructive and deformation field learning objectives. Additionally, we construct SPCV-based frameworks for both low-level and high-level 3D point cloud sequence processing and analysis tasks, including action recognition, temporal interpolation, and compression. Extensive experiments demonstrate the versatility and superiority of the proposed SPCV, which has the potential to offer new possibilities for deep learning on unstructured 3D point cloud sequences. Code will be released at https://github.com/ZENGYIMING-EAMON/SPCV.

Non-equilibrium magnetic properties of a mixed spin (1/2, 1) Ising graphene nanoisland.

Based on Glauber-type stochastic dynamics intertwined with mean field theory, we have investigated a graphene nanoisland's out-of-equilibrium magnetic and thermomagnetic behaviors with a mixed ferrimagnetic spin (1/2, 1) configuration in the presence of an oscillating magnetic field. The influence of the oscillating external magnetic field, single-ion anisotropy and temperature on the dynamic magnetic properties (dynamic phase transitions, hysteresis loop areas, correlations, and phase diagrams) of the system is studied in detail. The results illustrate that single ion anisotropy and the magnetic field play important roles in the system's dynamic order parameters, phase transitions and phase diagrams. Dynamic phase diagrams display some dynamic special points such as dynamic tricritical points, quadruple points and some sort of reentrant phenomena, which strongly depend on interaction parameters. Our predicted results may provide a reference for future experimental and theoretical studies of the nanostructures.

Dynamic Short Circuit Ratio for Stability Assessment of Grid Connected VSC-HVDC Systems Considering the Impact of Time-Varying Phase Shift

As more renewable resources are integrated into the receiving-end AC grid through VSC-HVDC, the grid connected VSC-HVDC system is facing a critical dynamic stability issue involving multiple-frequency oscillatory behaviors, which mainly stems from the low-system-strength and the time-varying phase shift. This paper proposes a general stability analysis method to measure the dynamic stability of grid connected VSC-HVDC system from the point of view of input to state stability (ISS). A nonlinear time-varying discrete-time state space model (NTDS <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> M) of the grid connected VSC-HVDC system is built considering the impact of the time-varying phase shift. Then, the NTDS <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> M is decomposed into several interconnected subsystems. The general asymptotic gain matrix (GAGM), whose elements are derived from the proposed dissipative-form sum-type inequalities (DSIs), is constructed to quantify the interactions between multiple subsystems. Finally, a general dynamic stability analysis index, called dynamic short circuit ratio (DSCR), is proposed to effectively explain the critical dynamic instability point of the grid connected VSC-HVDC system, based on the spectral radius property of the GAGM. Moreover, the real-time simulations and multiple comparative studies are implemented in this paper to verify the effectiveness of the proposed NTDS <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> M and stability analysis method by using RT-LAB platform.

Dynamic Line-Scan Point Cloud Correction Method Based on Two-Dimensional Image Reference

Dynamic Line-Scan Point Cloud Correction Method Based on Two-Dimensional Image Reference

Mixed-Precision Neural Architecture Search and Dynamic Split Point Selection for Split Computing

Mixed-Precision Neural Architecture Search and Dynamic Split Point Selection for Split Computing

Point source modelling approach for sessile droplet evaporation

Evaporation of sessile droplets from unheated solid surfaces is a ubiquitous process in many practical applications. A reduced order, analytical point source model (PSM) for the axisymmetric diffusion-dominated evaporation of an isolated sessile droplet surrounded by non-saturated, quiescent air was developed. The droplet is modeled as a dynamic point mass source in the limit of an isothermal system. The model also incorporates the spatial variation in the evaporative flux across the droplet free surface. The model is capable of considering the mode of evaporation, i.e., constant contact angle or contract radius. The PSM was simulated using the finite difference method in MATLAB R2020a. The model determines the vapor concentration distribution in the surrounding environment, the instantaneous evaporative flux averaged across the droplet surface and the overall evaporation rate. Calculating the evaporation rate assuming a spatially uniform evaporative flux under-predicts the evaporation rate by up to an order of magnitude. The model results agreed with experimental data in literature and sufficiently captures the evaporation process phenomena. The versatility and accurate predictive power of the PSM allows it to be a robust and computationally inexpensive modeling tool for studying sessile droplet evaporation in a wide range of technical applications.

Fluid dynamic parameters of naturally derived hydroxyapatite scaffolds for in vitro studies of bone cells

Hydroxyapatite scaffolds obtained from the biomorphic transformation of wood are characterized from a fluid dynamic point of view. Such material of recent introduction offers great advantages for the in vitro study of bone cells, mostly in virtue of its peculiar porous structure. Determining the flow resistance and morphological parameters of these scaffolds is an essential step towards their practical use in bioreactors and microfluidic devices. To this aim, a series of tests involving a draining fluid are performed on a set of disc-shaped scaffolds, followed by the microscopy analysis of the pores visible on the sample faces. Contrarily to what expected, a temperature dependence is observed for the flow resistance, even after normalizing it by the fluid properties. The interpretation of the experimental results is assisted by numerical outcomes from Computational Fluid Dynamics modelling, which underline some limitations in the application of classical laws to the present problem. While the complex and variable internal structure of the scaffolds prevents the systematic use of simplified formulae, a correlation is found between the flow resistance and the pore geometry, which can facilitate the characterization of further samples.

Adaptive LPU Decision for Dynamic Point Cloud Compression

Adaptive LPU Decision for Dynamic Point Cloud Compression

Low Complexity Coding Unit Decision for Video-Based Point Cloud Compression.

With growing demand for point cloud coding, Video-based Point Cloud Compression (V-PCC) is released for dynamic point clouds, relying on mature 2D video coding techniques. However, the huge computational complexity of 2D video codec is inherited by V-PCC, thereby resulting in a notably time-consuming encoding process for the projection videos. To accelerate the compression, this paper proposes a low complexity coding unit decision algorithm for V-PCC intra coding. First, the 2D sequences (occupancy, geometry, and attribute sequences) are projected from same 3D point could frames in V-PCC. By exploring the strong correlations among them, the cross-projection information is creatively proposed for improving the predication performance of CU partition. Second, considering the disparate coding losses generated by incorrect partitioning decisions of different CUs, we develop a rate-distortion-oriented learning approach aimed at increasing the decision accuracy of the CUs, severely affecting coding performance. Third, to accommodate the particular coding architecture of V-PCC intra configuration, we further devise an overall framework, including targeted feature extraction and partitioning decision for intra and inter coding of geometry and attribute sequences. The final experimental results strongly demonstrate the effectiveness of our proposed algorithm. The time consumption of total projection sequence compression can be reduced by 57.80%, while the coding losses on Geom.BD-TotalRate (D1 and D2) and Attr.BD-TotalRate Luma component are only 0.08%, 0.33%, and 0.16%, respectively, which can be negligible. To the best of our knowledge, the proposed algorithm achieves state-of-the-art performance for accelerating the projection sequence compression in V-PCC All-Intra configuration.

A Versatile Point Cloud Compressor Using Universal Multiscale Conditional Coding - Part I: Geometry.

A universal multiscale conditional coding framework, Unicorn, is proposed to compress the geometry and attribute of any given point cloud. Geometry compression is addressed in Part I of this paper, while attribute compression is discussed in Part II. We construct the multiscale sparse tensors of each voxelized point cloud frame and properly leverage lower-scale priors in the current and (previously processed) temporal reference frames to improve the conditional probability approximation or content-aware predictive reconstruction of geometry occupancy in compression. Unicorn is a versatile, learning-based solution capable of compressing static and dynamic point clouds with diverse source characteristics in both lossy and lossless modes. Following the same evaluation criteria, Unicorn significantly outperforms standard-compliant approaches like MPEG G-PCC, V-PCC, and other learning-based solutions, yielding state-of-the-art compression efficiency while presenting affordable complexity for practical implementations.

Inter-Frame Compression for Dynamic Point Cloud Geometry Coding.

Efficient point cloud compression is essential for applications like virtual and mixed reality, autonomous driving, and cultural heritage. This paper proposes a deep learning-based inter-frame encoding scheme for dynamic point cloud geometry compression. We propose a lossy geometry compression scheme that predicts the latent representation of the current frame using the previous frame by employing a novel feature space inter-prediction network. The proposed network utilizes sparse convolutions with hierarchical multiscale 3D feature learning to encode the current frame using the previous frame. The proposed method introduces a novel predictor network for motion compensation in the feature domain to map the latent representation of the previous frame to the coordinates of the current frame to predict the current frame's feature embedding. The framework transmits the residual of the predicted features and the actual features by compressing them using a learned probabilistic factorized entropy model. At the receiver, the decoder hierarchically reconstructs the current frame by progressively rescaling the feature embedding. The proposed framework is compared to the state-of-the-art Video-based Point Cloud Compression (V-PCC) and Geometry-based Point Cloud Compression (G-PCC) schemes standardized by the Moving Picture Experts Group (MPEG). The proposed method achieves more than 88% BD-Rate (Bjøntegaard Delta Rate) reduction against G-PCCv20 Octree, more than 56% BD-Rate savings against G-PCCv20 Trisoup, more than 62% BD-Rate reduction against V-PCC intra-frame encoding mode, and more than 52% BD-Rate savings against V-PCC P-frame-based inter-frame encoding mode using HEVC. These significant performance gains are cross-checked and verified in the MPEG working group.

Моделирование «ковидного провала» плановой госпитализации в период разгара пандемии COVID-19: анализ и оценка потерь использованной коечной мощности специализированного оториноларингологического стационара

Introduction. In the midst of the COVID-19 pandemic, healthcare organizations around the world were forced to change their operating hours, and some, especially those providing specialized medical care, were forced to stop operating completely. The purpose of the study is to develop a year-by-year model of the movement of patients before the pandemic and at its height, based on long-term data on the movement of inpatients. With its help, analyze possible losses in bed capacity (expressed in the number of idle bed days, or unserved patients) from the pandemic during 2020 — the year of the height of the pandemic. Materials and methods. The study was carried out according to the design of an observational retrospective analytical study. As research materials, we used anonymized data on the number of patients in the hospital at the Federal State Budgetary Institution «St. Petersburg Research Institute of ENT» of the Ministry of Health of Russia, obtained from the medical registration form N 007/u-02 „Sheet of daily records of the movement of patients and the bed capacity of a round-the-clock hospital, day hospital at a hospital facility“). The R and RStudio software environment was used as the main software product for solving statistical problems. Results. A load model has been created (the number of daily inpatients undergoing treatment) during the year. An analytical model of the “Covid failure” of bed capacity utilization during the height of the pandemic has been created. The coordinates of the dynamic points of bed release and return to typical bed occupancy values for this period of time were calculated (using linear regression forecasting). The share of hospital losses during the pandemic was calculated and assessed. Discussion. The use of “Covid failure” data superimposed on “pre-Covid” indicators allows not only to estimate annual losses, but also to predictively minimize losses during planned closures of hospital beds, for example, for repairs, given that the resumption of work of a medical institution’s beds will often be slower than the speed at which work stops. Analysis of linear regression coefficients of bed release and occupancy rates can give an idea of the speed of reaching standard values for the volume of medical care. As studies have shown during the COVID-19 pandemic, the minimum value of the quarantine period after the complete vacancy of beds is 14 days, and for high-risk scenarios 21 days. Based on this, for the conditions of organizing medical otorhinolaryngological care, the minimum predicted value of the time from the closure of hospitalization to its opening may be about 37 days after the start of patient discharge. The approach and mathematical functionality presented in the article can be used on the initial data of other institutions, as well as in preparation for the “Disease X” pandemic.

A kinetic model reveals the critical gating motifs for donor-substrate loading into Actinobacillus pleuropneumoniae N-glycosyltransferase.

Soluble N-glycosyltransferase from Actinobacillus pleuropneumoniae (ApNGT) catalyzes the glycosylation of asparagine residues, and represents one of the most encouraging biocatalysts for N-glycoprotein production. Since the sugar tolerance of ApNGT is restricted to limited monosaccharides (e.g., Glc, GlcN, Gal, Xyl, and Man), tremendous efforts are devoted to expanding the substrate scope of ApNGT via enzyme engineering. However, rational design of novel NGT variants suffers from an elusive understanding of the substrate-binding process from a dynamic point of view. Here, by employing extensive all-atom molecular dynamics (MD) simulations integrated with a kinetic model, we reveal, at the atomic level, the complete donor-substrate binding process from the bulk solvent to the ApNGT active-site, and the key intermediate states of UDP-Glc during its loading dynamics. We are able to determine the critical transition event that limits the overall binding rate, which guides us to pinpoint the key ApNGT residues dictating the donor-substrate entry. The functional roles of several identified gating residues were evaluated through site-directed mutagenesis and enzymatic assays. Two single-point mutations, N471A and S496A, could profoundly enhance the catalytic activity of ApNGT. Our work provides deep mechanistic insights into the structural dynamics of the donor-substrate loading process for ApNGT, which sets a rational basis for design of novel NGT variants with desired substrate specificity.

From address to outcome, a proposal for discussing research in the art academia towards the idea of a critical landscape

This text aims to discuss forms of teaching/learning that allow for the understanding of the involvement of students in carrying out actions that pertain to two major areas of intervention: landscape and knowledge, and how research processes may be generated by those actions. Landscape is intended to be approached from a dynamic and critical point of view, beyond its multiple senses and descriptive characters, such e.g. as rural or urban considered as limiteded descriptions. Knowledge is considered horizontally as a collectively generated process focused on providing tools for research and analysis based on student-centred actions. As a brief open-ended exercise, this text does not aim to respond to a set of challenges involved in the definition of the practices that will attempt to discuss, such as, firstly, the contradictions inherent in the definitions of trans or post-medial practices, in constant change and often contested from current theory and art itself; and second, the danger of enclosing ourselves in definitive terminologies to describe the practices that occupy us and that often operate precisely in opposition to the propensity to find and stabilise definitions, which is the aspiration of the academia. How is academic research in the art academia to deal with these contradictions and how to distinguish between practice based and practice led research, will be the key questions that the text will try to address critically. Is the space of the academia the last space for utopia? Cover page: Relational

Changes in boiling regime and heat transfer effectiveness during water droplet collision with a superheated wall

We experimentally investigated changes in boiling regime and heat transfer effectiveness when single droplets collided with superheated substrates at various initial temperatures. Saturated water droplets colliding at constant speeds were studied over a range of initial substrate temperatures that spanned all boiling regimes from nucleate through transition to film boiling. Space- and time-resolved infrared thermometry and convectional shadowgraph imaging were used to precisely derive local heat transfer distributions and detect characteristic boiling regimes. The results confirmed the existence of the typical regions of a boiling curve: nucleate, transition, and film. The transition boiling regime could be subdivided into three sub-regimes—bubbly, oscillating, and fingering—during which the heat transfer effectiveness (THE) varied in a non-linear manner, exhibiting local minima and maxima. To facilitate interpretation of non-linear variations in HTE during transition boiling, the basic physical parameters of local heat flux, contact area, and residence time were quantitatively measured using obtained thermometry images. In the transition boiling regime, HTE first rapidly decreased during bubbly boiling principally because of a substantial decrease in the droplet residence time. And THE increased during oscillating boiling because of contact area expansion and increased heat flux. Finally, THE decreased during fingering boiling due to substantial reduction in the area fraction of liquid contact. When liquid-solid contact was completely prohibited, stable film boiling was stable. Thus, the dynamic Leidenfrost temperature point could be detected based on heat transfer characteristics, rather than collision dynamics.

YES-SLAM: YOLOv7-enhanced-semantic visual SLAM for mobile robots in dynamic scenes

In dynamic scenes, moving objects will cause a significant error accumulation in robot’s pose estimation, and might even lead to tracking loss. In view of these problems, this paper proposes a semantic visual simultaneous localization and mapping algorithm based on YOLOv7. First, a light-weight network YOLOv7 is employed to acquire the semantic information of different objects in the scene, and flood filling and edge-enhanced techniques are combined to accurately and quickly separate the dynamic feature points from the extracted feature point set. In this way, the obtained static feature points with high-confidence are used to achieve the accurate estimation of robot’s pose. Then, according to the semantic information of YOLOv7, the motion magnitude of the robot, and the number of dynamic feature points in camera’s field-of-view, a high-performance keyframe selection strategy is constructed. On this basis, a robust loop closure detection method is developed by introducing the semantic information into the bag-of-words model, and global bundle adjustment optimization is performed on all keyframes and map points to obtain a global consistent pose graph. Finally, YOLOv7 is further utilized to carry out semantic segmentation on the keyframes, remove the dynamic objects in its semantic mask, and combine the point cloud pre-processing and octree map to build a 3D navigation semantic map. A series of simulations on TUM dataset and a case study in real scene clearly demonstrated the performance superiority of the proposed algorithms.

Numerical Simulation of Acoustic Emission Events in Supporting Roadways under Different Dynamic Loads

This paper studies the distribution characteristics of acoustic emission (AE) events in supporting roadways under different dynamic load conditions through numerical simulation and theoretical analysis. According to the principle of AE numerical simulation, the roadway model is established by FLAC3D software, and the supporting structure of the bolt cable is established by editing the Fish function. According to the model dynamic load application criterion, the relationship between the peak velocity of the particle, the distance from the source center to the dynamic impact damage point, and the dynamic load intensity is introduced. The velocity-distance-energy relationship is deduced from the relationship to determine the magnitude of the dynamic load energy. The simulation results show that the intensity of the dynamic load source and the distance of the dynamic load source directly impact the AE events. The larger the dynamic load intensity and the closer to the dynamic load source, the more concentrated the AE events. The roadway has a blocking effect on the transmission of the dynamic load stress waves. According to this characteristic, the roadway can be protected by a high-pressure relief roadway. Rock lithology greatly influences the transmission of dynamic load stress waves and the number of AE events. When the dynamic load stress wave passes through the rock strata of different lithologies, the attenuation of the dynamic load energy and the number of AE events are large. However, when the dynamic load stress wave passes through the rock strata of the same lithology, the attenuation of the dynamic load energy and AE events is small. The surrounding rock structure in the bottom corner area of the roadway is susceptible to disturbance from dynamic load sources above the roadway. The results are greatly significant for studying the AE characteristics of support roadways with disturbance-typeimpact failure.

DOE: a dynamic object elimination scheme based on geometric and semantic constraints

In this paper, we propose a dynamic object elimination algorithm that combines semantic and geometric constraints to address the problem of visual SLAM being easily affected by dynamic feature points in dynamic environments. This issue leads to the degradation of localisation accuracy and robustness. Firstly, we employ a lightweight YOLO-Tiny network to enhance both detection accuracy and system speed. Secondly, we integrate the YOLO-Tiny network into the ORB-SLAM3 system to extract semantic information from the images and initiate the elimination of dynamic feature points. Subsequently, we augment this approach by incorporating geometric constraints between neighbouring frames to further eliminate dynamic feature points. Then, the former is supplemented by combining the geometric constraints between neighbouring frames to further eliminate dynamic feature points. Experiments on the TUM dataset demonstrate that the algorithm in this paper can improve the Relative Pose Error (RPE) by up to 95.12% and the Absolute Trajectory Error (ATE) by up to 99.01% in high dynamic sequences compared to ORB-SLAM3. The effectiveness of dynamic feature point elimination is evident, leading to significantly improved localisation accuracy.