• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.7
• /
• pp.3599-3619
• /
• 2019

In human activity recognition system both static and motion information play crucial role for efficient and competitive results. Most of the existing methods are insufficient to extract video features and unable to investigate the level of contribution of both (Static and Motion) components. Our work highlights this problem and proposes Static-Motion fused features descriptor (SMFD), which intelligently leverages both static and motion features in the form of descriptor. First, static features are learned by two-stream 3D convolutional neural network. Second, trajectories are extracted by tracking key points and only those trajectories have been selected which are located in central region of the original video frame in order to to reduce irrelevant background trajectories as well computational complexity. Then, shape and motion descriptors are obtained along with key points by using SIFT flow. Next, cholesky transformation is introduced to fuse static and motion feature vectors to guarantee the equal contribution of all descriptors. Finally, Long Short-Term Memory (LSTM) network is utilized to discover long-term temporal dependencies and final prediction. To confirm the effectiveness of the proposed approach, extensive experiments have been conducted on three well-known datasets i.e. UCF101, HMDB51 and YouTube. Findings shows that the resulting recognition system is on par with state-of-the-art methods.

• Journal of Society of Preventive Korean Medicine
• /
• v.14 no.1
• /
• pp.13-24
• /
• 2010

Objectives : This study aimed to investigate the current job-related characteristics coordinators in East-West Collaborative Medical Practices(EWCMP) and to develop activation plans for them. Method : The participants were 51 personnel who were supporting EWCMP in the 28 institutions. Data were collected with self-administered questionnaires and analyzed with descriptive statistics. Results : The major role of the participants was educating and consulting patients(74.5%) and followed by supporting collaborating physician(70.6%). They assumed to be helpful to make the patients easy to use EWCMP(98%) and to give patients full information what they asked(96%). However, participants responded lack of adequate educational programs and role ambiguity as chief complaints(91.7%). They showed relatively high level of job importance(78.7points on 100points) and job satisfaction(72.8points on 100points). Coordinators were expected to have bachelor and more than 8 years clinical career, and communication skill. Conclusions : Coordinators have played key roles in giving information for the patients and coordinating EWCMP. However their roles and job description was not clear, and educational programs was insufficient as required. Therefore, it is needed to clarify their roles and job description and establish professional educational programs for supporting coordinators.

• The Journal of Korea Robotics Society
• /
• v.16 no.3
• /
• pp.179-188
• /
• 2021

Construction monitoring is one of the key modules in smart construction. Unlike structured urban environment, construction site mapping is challenging due to the characteristics of an unstructured environment. For example, irregular feature points and matching prohibit creating a map for management. To tackle this issue, we propose a system for data acquisition in unstructured environment and a framework for Intensity and Ambient Enhanced Lidar Inertial Odometry via Smoothing and Mapping, IA-LIO-SAM, that achieves highly accurate robot trajectories and mapping. IA-LIO-SAM utilizes a factor graph same as Tightly-coupled Lidar Inertial Odometry via Smoothing and Mapping (LIO-SAM). Enhancing the existing LIO-SAM, IA-LIO-SAM leverages point's intensity and ambient value to remove unnecessary feature points. These additional values also perform as a new factor of the K-Nearest Neighbor algorithm (KNN), allowing accurate comparisons between stored points and scanned points. The performance was verified in three different environments and compared with LIO-SAM.

• Journal of Korean Society for Quality Management
• /
• v.51 no.4
• /
• pp.677-689
• /
• 2023

Purpose: The purpose of this study is to design a x-ray screening rating system to improve X-ray screening ability, which is a core job competency of security screener at Incheon International Airport, and to verify its effectiveness through empirical analysis to suggest ways to improve the level management system. Methods: In this study, the effectiveness of the research model was analyzed using T-test tests for effect analysis based on the empirical analysis results derived through the competency evaluation model, the screening rating system. Results: The results of this study are as follows. The average score for regular education before the implementation of the x-ray screening rating system was 94.1 points, but after the implementation of the x-ray screening rating system, the average score for regular education was 95.5 points, an average of 1.4 points increased. In addition, the proportion of those with 95 or more points classified as high scorers also increased significantly from 51.1% to 69.3%. Conclusion: The X-ray screening rating system of security inspectors will systematically manage the level of screening ability, which is a key job competency, and play a strong role in improving competency, while preventing security accidents through early identification and intensive training of level-lowers.

• Nutrition Research and Practice
• /
• v.12 no.6
• /
• pp.459-468
• /
• 2018

BACKGROUND/OBJECTIVES: The Dietary Reference Intakes for Koreans (KDRIs), a set of reference intake values, have served as a basis for guiding a balanced diet that promotes health and prevents disease in the general Korean population. In the process of developing DRIs, a systematic review has played an important role in helping the DRI committees make evidence-based and transparent decisions for updating the next DRIs. Thus, the 2015 KDRI steering committee applied the systematic review framework to the revision process of the KDRIs. The purpose of this article is to summarize the revision process for the 2015 KDRIs by focusing on the systematic review framework. MATERIALS/METHODS: The methods used to develop the systematic review framework for 2015 KDRIs followed the Agency for Healthcare Research and Quality and the Tufts Evidence-based Practice Center. The framework for systematic review of the 2015 KDRIs comprised of the 3 following steps: (1) development of an analytic framework and refinement of key questions and search terms; (2) literature search and data extraction; and, (3) appraisal of the literature and summarizing the results. RESULTS: A total of 203,237 studies were retrieved through the above procedure, with 2,324 of these studies included in the analysis. General information, main results, comments of reviewers, and results of quality assessment were extracted and organized by study design. The average points of quality appraisals were 3.0 (range, 0-5) points for intervention, 6.1 (0-9) points for cohort, 6.0 (3-9) points for nested case-control, 5.4 (1-8) points for case-control, 14.6 (0-22) points for cross-sectional studies, and 7.0 (0-11) points for reviews. CONCLUSIONS: Systematic review helped to establish the 2015 KDRIs as a useful tool for evidence-based approach. Collaborative efforts to improve the framework for systematic review should be continued for future KDRIs.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.3
• /
• pp.65-70
• /
• 2019

The purpose of this study is to understand the college students' needs for software curriculum which based on surveys from educational satisfaction of the software lecture evaluation, as well as to find out the improvement plan by applying the text content analysis method. The research method used the text content analysis program to calculate the frequency of words occurrence, key words selection, co-occurrence frequency of key words, and analyzed the text center and network analysis by using the network analysis program. As a result of this research, the decent points of the software education network are mentioned with 'lecturer' is the most frequently occurrence after then with 'kindness', 'student', 'explanation', 'coding'. The network analysis of the shortage points has been the most mention of 'lecture', 'wish to', 'student', 'lecturer', 'assignment', 'coding', 'difficult', and 'announcement' which are mentioned together. The comprehensive network analysis of both good and shortage points has compared among key words, we can figure out difference among the key words: for example, 'group activity or task', 'assignment', 'difficulty on level of lecture', and 'thinking about lecturer'. Also, from this difference, we can provide that the lack of proper role of individual staff at group activities, difficult and excessive tasks, awareness of the difficulty and necessity of software education, lack of instructor's teaching method and feedback. Therefore, it is necessary to examine not only how the grouping of software education (activities) and giving assignments (or tasks), but also how carried out group activities and tasks and monitored about the contents of lectures, teaching methods, the ratio of practice and design thinking.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.14 no.2
• /
• pp.893-898
• /
• 2013

It is always a dream to recreate the experience of a particular place, the Panorama Virtual Reality has been interpreted as a kind of technology to create virtual environments and the ability to maneuver angle for and select the path of view in a dynamic scene. In this paper we examined an efficient method for Image registration and stitching of captured imaged. Two approaches are studied in this paper. First, dynamic programming is used to spot the ideal key points, match these points to merge adjacent images together, later image blending is used for smooth color transitions. In second approach, FAST and SURF detection are used to find distinct features in the images and nearest neighbor algorithm is used to match corresponding features, estimate homography with matched key points using RANSAC. The paper also covers the automatically choosing (recognizing, comparing) images to stitching method.

• Structural Engineering and Mechanics
• /
• v.56 no.4
• /
• pp.549-567
• /
• 2015

A first-order moment method (FORM) reliability analysis is commonly used for structural stability analysis. It requires the values and partial derivatives of the performance to function with respect to the random variables for the design. These calculations can be cumbersome when the performance functions are implicit. A Gaussian process (GP)-based response surface is adopted in this study to approximate the limit state function. By using a trained GP model, a large number of values and partial derivatives of the performance functions can be obtained for conventional reliability analysis with a FORM, thereby reducing the number of stability analysis calculations. This dynamic renewed knowledge source can provide great assistance in improving the predictive capacity of GP during the iterative process, particularly from the view of machine learning. An iterative algorithm is therefore proposed to improve the precision of GP approximation around the design point by constantly adding new design points to the initial training set. Examples are provided to illustrate the GP-based response surface for both structural and non-structural reliability analyses. The results show that the proposed approach is applicable to structural reliability analyses that involve implicit performance functions and structural response evaluations that entail time-consuming finite element analyses.

• Wind and Structures
• /
• v.20 no.5
• /
• pp.643-660
• /
• 2015

The wind tunnel test of large-scale sectional model and computational fluid dynamics (CFD) are employed for the purpose of studying the aerodynamic appendices and mechanism on suppression for the vortex-induced vibration (VIV). This paper takes the HongKong-Zhuhai-Macao Bridge as an example to conduct the wind tunnel test of large-scale sectional model. The results of wind tunnel test show that it is the crash barrier that induces the vertical VIV. CFD numerical simulation results show that the distance between the curb and crash barrier is not long enough to accelerate the flow velocity between them, resulting in an approximate stagnation region forming behind those two, where the continuous vortex-shedding occurs, giving rise to the vertical VIV in the end. According to the above, 3 types of wind fairing (trapezoidal, airfoil and smaller airfoil) are proposed to accelerate the flow velocity between the crash barrier and curb in order to avoid the continuous vortex-shedding. Both of the CFD numerical simulation and the velocity field measurement show that the flow velocity of all the measuring points in case of the section with airfoil wind fairing, can be increased greatly compared to the results of original section, and the energy is reduced considerably at the natural frequency, indicating that the wind fairing do accelerate the flow velocity behind the crash barrier. Wind tunnel tests in case of the sections with three different countermeasures mentioned above are conducted and the results compared with the original section show that all the three different countermeasures can be used to control VIV to varying degrees.

• Journal of the Korea Computer Graphics Society
• /
• v.29 no.1
• /
• pp.1-11
• /
• 2023

Deep-learning-based image segmentation is one of the most widely employed lane detection approaches, and it requires a post-process for extracting the key points on the lanes. A general approach for key-point extraction is using a fixed threshold defined by a user. However, finding the best threshold is a manual process requiring much effort, and the best one can differ depending on the target data set (or an image). We propose a novel key-point extraction algorithm that automatically adapts to the target image without any manual threshold setting. In our adaptive key-point extraction algorithm, we propose a line-level normalization method to distinguish the lane region from the background clearly. Then, we extract a representative key point for each lane at a line (row of an image) using a kernel density estimation. To check the benefits of our approach, we applied our method to two lane-detection data sets, including TuSimple and CULane. As a result, our method achieved up to 1.80%p and 17.27% better results than using a fixed threshold in the perspectives of accuracy and distance error between the ground truth key-point and the predicted point.