scholarly journals Ensemble Spatio-Temporal Distance Net for Skeleton Based Action Recognition

2019 ◽  
Vol 20 (3) ◽  
pp. 485-494
Author(s):  
M Naveenkumar ◽  
S Domnic
Keyword(s):  
Action Recognition ◽  
Short Term Memory ◽  
Temporal Dynamics ◽  
Spatial Dynamics ◽  
Recognition System ◽  
Temporal Distance ◽  
Spatial Distance ◽  
Spatial And Temporal Dynamics ◽  
Entire Sequence ◽  
Spatio Temporal

The performance of an efficient and accurate action recognition system heavily depends on distinctive representations for a different class of action sequences. To address this issue, we propose an ensemble network in this paper. We design two multilayer Long Short Term Memory networks to capture spatial and temporal dynamics of the entire sequence, referred to as Spatial-distance Net (SdNet) and Temporal-distance Net (TdNet) respectively. More specifically, SdNet captures the spatial dynamics of joints within a frame and TdNet explores the temporal dynamics of joints between frames along the sequence. Finally, two nets are fused as one Ensemble network, referred to as Spatio -Temporal distance Net (STdNet) to explore both spatial and temporal dynamics. The efficacy of the proposed method is evaluated on two widely used datasets, UTD MHAD and NTU RGB+D, and the proposed STdNet achieved 91.16% and 80.03% accuracies respectively.

Spatial and temporal dynamics of Pacific capelin Mallotus catervarius in the Gulf of Alaska: implications for ecosystem-based fisheries management

2020 ◽  
Vol 637 ◽  
pp. 117-140 ◽  
Author(s):  
DW McGowan ◽  
ED Goldstein ◽  
ML Arimitsu ◽  
AL Deary ◽  
O Ormseth ◽  
...  
Keyword(s):  
Temporal Dynamics ◽  
Marine Ecosystem ◽  
Gulf Of Alaska ◽  
Data Series ◽  
Limited Information ◽  
Important Species ◽  
Multiple Data ◽  
Spatial And Temporal Dynamics ◽  
Spawning Areas ◽  
Spatio Temporal

Pacific capelin Mallotus catervarius are planktivorous small pelagic fish that serve an intermediate trophic role in marine food webs. Due to the lack of a directed fishery or monitoring of capelin in the Northeast Pacific, limited information is available on their distribution and abundance, and how spatio-temporal fluctuations in capelin density affect their availability as prey. To provide information on life history, spatial patterns, and population dynamics of capelin in the Gulf of Alaska (GOA), we modeled distributions of spawning habitat and larval dispersal, and synthesized spatially indexed data from multiple independent sources from 1996 to 2016. Potential capelin spawning areas were broadly distributed across the GOA. Models of larval drift show the GOA’s advective circulation patterns disperse capelin larvae over the continental shelf and upper slope, indicating potential connections between spawning areas and observed offshore distributions that are influenced by the location and timing of spawning. Spatial overlap in composite distributions of larval and age-1+ fish was used to identify core areas where capelin consistently occur and concentrate. Capelin primarily occupy shelf waters near the Kodiak Archipelago, and are patchily distributed across the GOA shelf and inshore waters. Interannual variations in abundance along with spatio-temporal differences in density indicate that the availability of capelin to predators and monitoring surveys is highly variable in the GOA. We demonstrate that the limitations of individual data series can be compensated for by integrating multiple data sources to monitor fluctuations in distributions and abundance trends of an ecologically important species across a large marine ecosystem.

MDTP

2021 ◽  
Vol 14 (8) ◽  
pp. 1289-1297
Author(s):  
Ziquan Fang ◽  
Lu Pan ◽  
Lu Chen ◽  
Yuntao Du ◽  
Yunjun Gao
Keyword(s):  
Neural Network ◽  
Short Term Memory ◽  
Temporal Dynamics ◽  
Real Life ◽  
Feature Modeling ◽  
Traffic Prediction ◽  
Interactive System ◽  
Trajectory Data ◽  
Spatio Temporal ◽  
Prediction Approach

Traffic prediction has drawn increasing attention for its ubiquitous real-life applications in traffic management, urban computing, public safety, and so on. Recently, the availability of massive trajectory data and the success of deep learning motivate a plethora of deep traffic prediction studies. However, the existing neural-network-based approaches tend to ignore the correlations between multiple types of moving objects located in the same spatio-temporal traffic area, which is suboptimal for traffic prediction analytics. In this paper, we propose a multi-source deep traffic prediction framework over spatio-temporal trajectory data, termed as MDTP. The framework includes two phases: spatio-temporal feature modeling and multi-source bridging. We present an enhanced graph convolutional network (GCN) model combined with long short-term memory network (LSTM) to capture the spatial dependencies and temporal dynamics of traffic in the feature modeling phase. In the multi-source bridging phase, we propose two methods, Sum and Concat, to connect the learned features from different trajectory data sources. Extensive experiments on two real-life datasets show that MDTP i) has superior efficiency, compared with classical time-series methods, machine learning methods, and state-of-the-art neural-network-based approaches; ii) offers a significant performance improvement over the single-source traffic prediction approach; and iii) performs traffic predictions in seconds even on tens of millions of trajectory data. we develop MDTP + , a user-friendly interactive system to demonstrate traffic prediction analysis.

scholarly journals ON MEMORY AND STRUCTURAL DYNAMISM IN EXCITABLE CELLULAR AUTOMATA WITH DEFENSIVE INHIBITION

2008 ◽  
Vol 18 (02) ◽  
pp. 527-539 ◽  
Author(s):  
RAMÓN ALONSO-SANZ ◽  
ANDREW ADAMATZKY
Keyword(s):  
Cellular Automata ◽  
Short Term Memory ◽  
Temporal Dynamics ◽  
Short Term ◽  
Resting Cell ◽  
Fixed Topology ◽  
Dynamical Topology ◽  
Novel Complex ◽  
Spatio Temporal ◽  
Update Rules

Commonly studied cellular automata are memoryless and have fixed topology of connections between cells. However by allowing updates of links and short-term memory in cells we may potentially discover novel complex regimes of spatio-temporal dynamics. Moreover, by adding memory and dynamical topology to state update rules we somehow forge elementary but nontraditional models of neurons networks (aka neuron layers in frontal parts). In the present paper, we demonstrate how this can be done on a self-inhibitory excitable cellular automata. These automata imitate a phenomenon of inhibition caused by hight-strength stimulus: a resting cell excites if there are one or two excited neighbors, the cell remains resting otherwise. We modify the automaton by allowing cells to have few-steps memories, and create links between neighboring cells removed or generated depending on the states of the cells.

scholarly journals Weevil borers affect the spatio-temporal dynamics of banana Fusarium wilt

2021 ◽  
Author(s):  
Daniel Heck ◽  
Gabriel Alves ◽  
Eduardo S. G. Mizubuti
Keyword(s):  
Fusarium Wilt ◽  
Temporal Dynamics ◽  
Insect Pests ◽  
Field Conditions ◽  
Disease Progress ◽  
Spatial And Temporal Dynamics ◽  
Comparative Epidemiology ◽  
Progress Rate ◽  
Spatio Temporal ◽  
Metamasius Hemipterus

AbstractDispersal of propagules of a pathogen has remarkable effects on the development of epidemics. Previous studies suggested that insect pests play a role in the development of Fusarium wilt (FW) epidemics in banana fields. We provided complementary evidence for the involvement of two insect pests of banana, the weevil borer (Cosmopolites sordidus L. - WB) and the false weevil borer (Metamasius hemipterus L. - FWB), in the dispersal of Fusarium oxysporum f. sp. cubense (Foc) using a comparative epidemiology approach under field conditions. Two banana plots located in a field with historical records of FW epidemics were used, one was managed with Beauveria bassiana to reduce the population of weevils, and the other was left without B. bassiana applications. The number of WB and FWB was monitored biweekly and the FW incidence was quantified bimonthly during two years. The population of WB and the incidence (6.7%) of FW in the plot managed with B. bassiana were lower than in the plot left unmanaged (13%). The monomolecular model best fitted the FW disease progress data and, as expected, the average estimated disease progress rate was lower in the plot managed with the entomopathogenic fungus (r = 0.0024) compared to the unmanaged plot (r = 0.0056). Aggregation of FW was higher in the field with WB management. WB affected the spatial and temporal dynamics of FW epidemics under field conditions and brought evidence that managing the insects may reduce FW of bananas intensity.

A Flexible High-Level Fusion for an Accurate Human Action Recognition System

2020 ◽  
Vol 29 (12) ◽  
pp. 2050190
Author(s):  
Amel Ben Mahjoub ◽  
Mohamed Atri
Keyword(s):  
Action Recognition ◽  
Short Term Memory ◽  
Recognition Rate ◽  
Human Action Recognition ◽  
Local Binary Patterns ◽  
Low Complexity ◽  
Recognition System ◽  
Human Action ◽  
Collaborative Representation ◽  
Depth Video

Action recognition is a very effective method of computer vision areas. In the last few years, there has been a growing interest in Deep learning networks as the Long Short–Term Memory (LSTM) architectures due to their efficiency in long-term time sequence processing. In the light of these recent events in deep neural networks, there is now considerable concern about the development of an accurate action recognition approach with low complexity. This paper aims to introduce a method for learning depth activity videos based on the LSTM and the classification fusion. The first step consists in extracting compact depth video features. We start with the calculation of Depth Motion Maps (DMM) from each sequence. Then we encode and concatenate contour and texture DMM characteristics using the histogram-of-oriented-gradient and local-binary-patterns descriptors. The second step is the depth video classification based on the naive Bayes fusion approach. Training three classifiers, which are the collaborative representation classifier, the kernel-based extreme learning machine and the LSTM, is done separately to get classification scores. Finally, we fuse the classification score outputs of all classifiers with the naive Bayesian method to get a final predicted label. Our proposed method achieves a significant improvement in the recognition rate compared to previous work that has used Kinect v2 and UTD-MHAD human action datasets.

scholarly journals Methodological Proposal for the Study of Temporal and Spatial Dynamics during the Late Period between the Middle Ebro and the Pyrenees

Information ◽  
2020 ◽  
Vol 11 (7) ◽  
pp. 366 ◽  
Author(s):  
Leticia Tobalina Pulido
Keyword(s):  
Spatial Databases ◽  
Temporal Dynamics ◽  
Methodological Approach ◽  
Spatial Dynamics ◽  
Descriptive Statistics ◽  
Late Period ◽  
Rural Settlements ◽  
Spatial And Temporal Dynamics ◽  
Temporal And Spatial ◽  
Space Occupation

The article I present here deals with the methodological approach carried out in my PhD in which I analyzed the spatial and temporal dynamics of late rural settlements during five centuries in the southern Pyrenees area, using geographic information systems, spatial databases, and descriptive statistics to establish models of space occupation and try to determine how these vary over the different centuries.

scholarly journals Modeling Spatio-temporal Dynamics of Chromatin Marks

2017 ◽  
Author(s):  
Petko Fiziev ◽  
Jason Ernst
Keyword(s):  
Gene Expression ◽  
Time Course ◽  
Temporal Dynamics ◽  
Spatial Dynamics ◽  
Computational Method ◽  
Time Points ◽  
Signal Density ◽  
Spatial Changes ◽  
Spatio Temporal ◽  
Over Time

ABSTRACTTo model spatial changes of chromatin mark peaks over time we developed and applied ChromTime, a computational method that predicts regions for which peaks either expand or contract significantly or hold steady between time points. Predicted expanding and contracting peaks can mark regulatory regions associated with transcription factor binding and gene expression changes. Spatial dynamics of peaks provided information about gene expression changes beyond localized signal density changes. ChromTime detected asymmetric expansions and contractions, which for some marks associated with the direction of transcription. ChromTime facilitates the analysis of time course chromatin data in a range of biological systems.

scholarly journals Spatio-Temporal Dynamics of Municipal Planning, a case study of Barahathawa Municipality, Nepal

2021 ◽  
Vol 4 (2) ◽  
Author(s):  
Kedar Dahal ◽  
Krishna P. Timilsina
Keyword(s):  
Urban Growth ◽  
Agricultural Land ◽  
Temporal Dynamics ◽  
Transportation Network ◽  
Urban Expansion ◽  
Infrastructure Development ◽  
Rural Settlements ◽  
Goods And Services ◽  
Spatial And Temporal Dynamics ◽  
Spatio Temporal

The Rapid transformation of rural settlements into municipalities in Nepal has brought significant changes in land use and urban expansion patterns mostly through the conversion of agricultural land into the built-up area. The issue is studied taking a case of rapidly growing town Barahathawa Municipality of Sarlahi District. After the declaration of the municipality, several new roads have been opened and upgraded; and the municipality has well-connected to the national transportation network. After promulgated the Constitution of Nepal 2015 and elected local bodies, the municipality budget has been increased significantly as a result of increasing municipal investment in socio-economic and physical infrastructure development and environmental protection which have attracted people, goods, and services creating the zone of influence. One of the changes found in the municipality is the increasing built-up area and expansion of urban growth through the decreasing agricultural land. Urban growth has been observed taking place around the Barahathawa Bazaar and main roadsides. The built-up area in Barahathawa municipality has remarkably increased by 184% with the decrease of shrub and agricultural land within 10 years. Implications of such spatial and temporal dynamics have been a core issue of urban planning in most of the newly declared municipalities in Nepal

scholarly journals Assessing the Wall-To-Wall Spatial and Qualitative Dynamics of the Brazilian Pasturelands, between 2010 and 2018, Based on the Analysis of the Landsat Data Archive

2021 ◽  
Author(s):  
Claudinei Oliveira-Santos ◽  
Vinicius Vieira Mesquita ◽  
Leandro Leal Parente ◽  
Alexandre de Siqueira Pinto ◽  
Laerte Guimaraes Ferreira
Keyword(s):  
Temporal Dynamics ◽  
Spatial Dynamics ◽  
Limited Resource ◽  
Change Scenario ◽  
The North ◽  
Pasture Quality ◽  
Spatio Temporal ◽  
Pasture Area ◽  
Degraded Pastures

The Brazilian livestock is predominantly extensive, with approximately 90% of the production being sustained on pasture, which occupies around 20% of the territory. In the current climate change scenario and where cropland is becoming a limited resource, there is a growing need for a more efficient land use and occupation. It is estimated that more than half of the Brazilian pastures have some level of degradation; however there is still no mapping of the quality of pastures on a national scale. In this study, we mapped and evaluated the spatio-temporal dynamics of pasture quality in Brazil, between 2010 and 2018, considering three classes of degradation: Absent (D0), Intermediate (D1), and Severe (D2). There was no variation in the total area occupied by pastures in the evaluated period, in spite of the accentuated spatial dynamics, with a retraction in the center-south and expansion to the north, over areas of ​​native vegetation. The percentage of non-degraded pastures increased ~12%, due to the recovery of degraded areas and the emergence of new pasture areas as a result of the prevailing spatial dynamics. However, about 44 Mha of the pasture area is currently severely degraded. The dynamics in pasture quality were not homogeneous in property size classes. We observed that in the approximately 2.68 million properties with livestock activity, the proportion with quality gains was twice as low in small properties compared to large ones, and the proportion with losses was three times greater, showing an increase in inequality between properties with more and less resources (large and small, respectively). The areas occupied by pastures in Brazil present an unique opportunity to increase livestock production and make available areas for agriculture, without the need for new deforestation in the coming decades.

Dual-Stream Structured Graph Convolution Network for Skeleton-Based Action Recognition

2021 ◽  
Vol 17 (4) ◽  
pp. 1-22
Author(s):  
Chunyan Xu ◽  
Rong Liu ◽  
Tong Zhang ◽  
Zhen Cui ◽  
Jian Yang ◽  
...  
Keyword(s):  
Action Recognition ◽  
Temporal Dynamics ◽  
Recognition Task ◽  
Human Movement ◽  
Body Part ◽  
Human Action ◽  
Human Motion ◽  
Body Parts ◽  
Dynamic Interactions ◽  
Spatio Temporal

In this work, we propose a dual-stream structured graph convolution network ( DS-SGCN ) to solve the skeleton-based action recognition problem. The spatio-temporal coordinates and appearance contexts of the skeletal joints are jointly integrated into the graph convolution learning process on both the video and skeleton modalities. To effectively represent the skeletal graph of discrete joints, we create a structured graph convolution module specifically designed to encode partitioned body parts along with their dynamic interactions in the spatio-temporal sequence. In more detail, we build a set of structured intra-part graphs, each of which can be adopted to represent a distinctive body part (e.g., left arm, right leg, head). The inter-part graph is then constructed to model the dynamic interactions across different body parts; here each node corresponds to an intra-part graph built above, while an edge between two nodes is used to express these internal relationships of human movement. We implement the graph convolution learning on both intra- and inter-part graphs in order to obtain the inherent characteristics and dynamic interactions, respectively, of human action. After integrating the intra- and inter-levels of spatial context/coordinate cues, a convolution filtering process is conducted on time slices to capture these temporal dynamics of human motion. Finally, we fuse two streams of graph convolution responses in order to predict the category information of human action in an end-to-end fashion. Comprehensive experiments on five single/multi-modal benchmark datasets (including NTU RGB+D 60, NTU RGB+D 120, MSR-Daily 3D, N-UCLA, and HDM05) demonstrate that the proposed DS-SGCN framework achieves encouraging performance on the skeleton-based action recognition task.

Sign in / Sign up

Export Citation Format

Share Document