• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.14 no.4
• /
• pp.314-322
• /
• 2021

Skeleton-based action recognition has attracted considerable attention in human action recognition. Recent methods for skeleton-based action recognition employ spatiotemporal graph convolutional networks (GCNs) and have remarkable performance. However, most of them have heavy computational complexity for robust action recognition. To solve this problem, we propose a shuffle graph convolutional network (SGCN) which is a lightweight graph convolutional network using pointwise group convolution rather than pointwise convolution to reduce computational cost. Our SGCN is composed of spatial and temporal GCN. The spatial shuffle GCN contains pointwise group convolution and part shuffle module which enhances local and global information between correlated joints. In addition, the temporal shuffle GCN contains depthwise convolution to maintain a large receptive field. Our model achieves comparable performance with lowest computational cost and exceeds the performance of baseline at 0.3% and 1.2% on NTU RGB+D and NTU RGB+D 120 datasets, respectively.

• BMB Reports
• /
• v.55 no.8
• /
• pp.363-369
• /
• 2022

Biomolecular phase separation has recently attracted broad interest, due to its role in the spatiotemporal compartmentalization of living cells. It governs the formation, regulation, and dissociation of biomolecular condensates, which play multiple roles in vivo, from activating specific biochemical reactions to organizing chromatin. Interestingly, biomolecular phase separation seems to be a mainly passive process, which can be explained by relatively simple physical principles and reproduced in vitro with a minimal set of components. This Mini review focuses on our current understanding of the fundamental principles of biomolecular phase separation and the recent progress in the research on this topic.

• Medical Lasers
• /
• v.9 no.1
• /
• pp.6-11
• /
• 2020

Hydrogels have been developed and used in tissue engineering and regenerative medicine to deliver therapeutics to injured or diseased tissue because of their versatility and properties that can be tailored to match the natural extracellular matrix. Hydrogels can be made with a variety of physical and chemical properties combined with light responsiveness ideal for applications in different fields of medicine that require the spatiotemporal control of therapeutics. Light, as a stimulus, is relatively inexpensive, contact-free, noninvasive with high spatial resolution and temporal control, convenient and easy to use, and allows deep tissue penetration that is relatively harmless. Photoresponsive hydrogels are ideal candidates for on-demand drug delivery systems that are capable of sustained and controlled drug release, minimizing the side effects, and ensuring the activity and efficient delivery of drugs to the target tissue.

• ETRI Journal
• /
• v.44 no.2
• /
• pp.327-338
• /
• 2022

Classification models for human action recognition require robust features and large training sets for good generalization. However, data augmentation methods are employed for imbalanced training sets to achieve higher accuracy. These samples generated using data augmentation only reflect existing samples within the training set, their feature representations are less diverse and hence, contribute to less precise classification. This paper presents new data augmentation and action representation approaches to grow training sets. The proposed approach is based on two fundamental concepts: virtual video generation for augmentation and representation of the action videos through robust features. Virtual videos are generated from the motion history templates of action videos, which are convolved using a convolutional neural network, to generate deep features. Furthermore, by observing an objective function of the genetic algorithm, the spatiotemporal features of different samples are combined, to generate the representations of the virtual videos and then classified through an extreme learning machine classifier on MuHAVi-Uncut, iXMAS, and IAVID-1 datasets.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.732-735
• /
• 2020

기존에 지도학습 방법은 성능은 좋지만, 학습할 때 비디오 데이터와 정답 라벨이 있어야 한다. 그러나 이러한 데이터의 라벨을 수동으로 붙여줘야 하는 문제점과 그에 필요한 시간과 돈이 크다는 것이다. 이러한 문제점을 해결하기 위한 다양한 방법 중 자기지도학습(Self-Supervised Learning) 중 하나인 회전 방법을 비디오 데이터에 적용하여 학습하는 연구를 진행하였다. 본 연구에서는 두가지 방법을 제안한다. 먼저 기존의 비디오 데이터를 입력으로 받으면 단순히 비디오 자체를 회전시키는 것이 아닌 입력으로 들어온 비디오의 각각 프레임이 시간이 지나면서 일정한 속도로 회전을 시킨다. 이때의 회전은 총 네 가지 각도[0, 90, 180, 270]를 분류하도록 하는 방법론이다. 두 번째로 비디오의 프레임이 시간이 지나면서 변할 때 프레임 별로 고정된 각도로 회전시키는데 이때 회전하는 속도 네 가지 [1x, 0.5x, 0.25x, 0.125]를 분류하도록 하는 방법론이다. 이와 같은 제안하는 pretext task들을 통해 네트워크를 학습한 뒤, 학습된 모델을 fine tune 시켜 비디오 분류에 대한 실험을 수행 및 결과를 도출하였다.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.1
• /
• pp.58.4-59
• /
• 2021

This presentation introduces Korea's SNIPE (Small scale magNespheric and Ionospheric Plasma Experiment) mission, formation flying CubeSat constellation. Observing particles and waves on a single satellite suffers from inherent space-time ambiguity. To observe spatial and temporal variations of the micro-scale plasma structures on the topside ionosphere, four 6U CubeSats (~ 10 kg) will be launched into a polar orbit of the altitude of ~500 km in 2021. The distances of each satellite will be controlled from 10 km to more than 100 km by formation flying algorithm. The SNIPE mission is equipped with identical scientific instruments, solid-state telescope, magnetometer, and Langmuir probe. All the payloads have a high temporal resolution (sampling rates of about 10 Hz). Iridium modules provide an opportunity to upload changes in operational modes when geomagnetic storms occur. SNIPE's observations of the dimensions, occurrence rates, amplitudes, and spatiotemporal evolution of polar cap patches, field-aligned currents (FAC), radiation belt microbursts, and equatorial and mid-latitude plasma blobs and bubbles will determine their significance to the solar wind-magnetosphere-ionosphere interaction and quantify their impact on space weather.

• Asia pacific journal of information systems
• /
• v.33 no.2
• /
• pp.318-341
• /
• 2023

Mobile TV refers to the service that provides live broadcasting and video-on-demand content through a mobile device. In addition to the advertisement as the early-stage revenue model, the paid subscription model has emerged as a more sustainable revenue source for mobile TV services. In this study, with the surveys of 450 free mobile TV users, we examine the motivational factors influencing their intention to adopt a paid subscription model. Results show that three extrinsic motivations, price fairness, subjective norm, and mobile TV utilization, are positively associated with free users' paid subscription intention. In contrast, intrinsic motivations, such as hedonic need, spatiotemporal convenience, and self-efficacy, have no significant influence on the intention. We also found that the expected value is positively associated with attitude toward mobile TV service, also positively influencing the paid subscription intention.

• Wind and Structures
• /
• v.38 no.4
• /
• pp.295-307
• /
• 2024

The funnel-shaped vortex structure of tornadoes results in a spatiotemporally varying wind velocity (speed and direction) field. However, very limited full-scale tornado data along the height and radius positions are available to identify and reliably establish a description of complex vortex structure together with the resulting aerodynamic effects on the high-speed train (HST). In this study, the improved delayed detached eddy simulation (IDDES) for flow structures and aerodynamic pressures around an HST under tornado-like winds are conducted to provide high-fidelity computational fluid dynamics (CFD) results. To demonstrate the accuracy of the numerical method adopted in this study, both field observations and wind-tunnel data are utilized to respectively validate the simulated tornado flow fields and HST aerodynamics. Then, the flow structures and aerodynamic pressures (as well as aerodynamic forces and moments) around the HST at various locations within the tornado-like vortex are comprehensively compared to highlight the importance of considering the complex spatiotemporal wind features in the HST-tornado interactions.

• Journal of the Korean Society for Marine Environment & Energy
• /
• v.11 no.2
• /
• pp.78-85
• /
• 2008

We estimated the pollutant loads for the last 3 years based on the daily discharge at the Nakdong River dam(barrage) and spatiotemporal characteristics of seawater quality in the Nakdong river estuary to investigate the correlation between the pollutant load inflow rate and seawater quality. The main results from this research are summarized as follows. (1) The total discharge at the Nakdong River dam dam the last 11 years has been $224,576.8{\times}10^6m^3/day$. The discharge figures show that the maximum discharge occurs in August with $52,634.2{\times}10^6 m^3/day$ (23.4% of the year's volume), followed by July and Sep. in that order with 23.1 and 17%, respectively. (2) The pollutant load influx from the Nakdong River dam was composed of 307,591.3COD-kg/day, 128.746.1 TN-kg/day, and 107,625.8 TP-kg/day. (3) The surface temperature in the Nakdong River estuary was about $2.137^{\circ}C$ higher than that of the lower layer. The salinity of the lower layer was 2.209%o higher than that of the ocean surface. The salinity of the ocean surface decreased by up to 19.593%o due to the inflow of the discharge at the Nakdong River dam. (4) DO, COD, TN, and SS concentration levels tended to be higher at the ocean surface than in lower layers, whereas the reverse was true for TP. (5) The water mass at the ocean's surface and in the lower layers during the drought and flood seasons tended to be separated by the difference in densities due to the freshwater inflow.

• Journal of Cadastre & Land InformatiX
• /
• v.47 no.2
• /
• pp.233-254
• /
• 2017

The purpose of this study is to analyze the main factors influencing the severity of traffic accidents and to visualize spatiotemporal characteristics of traffic accidents in Seoul. To do this, we collected the traffic accident data that occurred in Seoul for four years from 2012 to 2015, and classified as slight, serious, and death traffic accidents according to the severity of traffic accidents. The analysis of spatiotemporal characteristics of traffic accidents was performed by kernel density analysis, hotspot analysis, space time cube analysis, and Emerging HotSpot Analysis. The factors affecting the severity of traffic accidents were analyzed using decision tree model. The results show that traffic accidents in Seoul are more frequent in suburbs than in central areas. Especially, traffic accidents concentrated in some commercial and entertainment areas in Seocho and Gangnam, and the traffic accidents were more and more intense over time. In the case of death traffic accidents, there were statistically significant hotspot areas in Yeongdeungpo-gu, Guro-gu, Jongno-gu, Jung-gu and Seongbuk. However, hotspots of death traffic accidents by time zone resulted in different patterns. In terms of traffic accident severity, the type of accident is the most important factor. The type of the road, the type of the vehicle, the time of the traffic accident, and the type of the violation of the regulations were ranked in order of importance. Regarding decision rules that cause serious traffic accidents, in case of van or truck, there is a high probability that a serious traffic accident will occur at a place where the width of the road is wide and the vehicle speed is high. In case of bicycle, car, motorcycle or the others there is a high probability that a serious traffic accident will occur under the same circumstances in the dawn time.