• Journal of Korea Multimedia Society
• /
• v.24 no.5
• /
• pp.659-666
• /
• 2021

With the continuous development of deep learning, human behavior recognition algorithms have achieved good results. However, in a multi-person recognition environment, the complex behavior environment poses a great challenge to the efficiency of recognition. To this end, this paper proposes a multi-person pose estimation model. First of all, the human detectors in the top-down framework mostly use the two-stage target detection model, which runs slow down. The single-stage YOLOv3 target detection model is used to effectively improve the running speed and the generalization of the model. Depth separable convolution, which further improves the speed of target detection and improves the model's ability to extract target proposed regions; Secondly, based on the feature pyramid network combined with context semantic information in the pose estimation model, the OHEM algorithm is used to solve difficult key point detection problems, and the accuracy of multi-person pose estimation is improved; Finally, the Euclidean distance is used to calculate the spatial distance between key points, to determine the similarity of postures in the frame, and to eliminate redundant postures.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.5
• /
• pp.801-808
• /
• 2022

Research on how to embed knowledge in large-scale Linked Data and apply neural network models for entity matching is relatively scarce. The most fundamental problem with this is that different labels lead to lexical heterogeneity. In this paper, we propose an extended GCN (Graph Convolutional Network) model that combines re-align structure to solve this lexical heterogeneity problem. The proposed model improved the performance by 53% and 40%, respectively, compared to the existing embedded-based MTransE and BootEA models, and improved the performance by 5.1% compared to the GCN-based RDGCN model.

• Proceedings of the IEEK Conference
• /
• 2000.11d
• /
• pp.159-162
• /
• 2000

In this paper, we propose an efficient method to detect shot changes in compressed MPEG video data by using reference features among video frames. The reference features among video frames imply the similarities among adjacent frames by prediction coded type of each frame. A shot change is detected if the similarity degrees of a frame and its adjacent frames are low. And the shot change detection algorithm is improved by using Fuzzy c-means (FCM) clustering algorithm. The FCM clustering algorithm uses the shot change probabilities evaluated in the mask matching of reference ratios and difference measure values based on frame reference ratios.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.6
• /
• pp.2748-2766
• /
• 2016

Compared with traditional patch-based sparse representation, recent studies have concluded that group-based sparse representation (GSR) can simultaneously enforce the intrinsic local sparsity and nonlocal self-similarity of images within a unified framework. This article investigates an accelerated split Bregman method (SBM) that is based on GSR which exploits image compressive sensing (CS). The computational efficiency of accelerated SBM for the measurement matrix of a partial Fourier matrix can be further improved by the introduction of a fast Fourier transform (FFT) to derive the enhanced algorithm. In addition, we provide convergence analysis for the proposed method. Experimental results demonstrate that accelerated SBM is potentially faster than some existing image CS reconstruction methods.

• Journal of The Korean Astronomical Society
• /
• v.36 no.spc1
• /
• pp.1-5
• /
• 2003

In an early paper Skumanich suggested the existence of a scaling law relating the mean sunspot magnetic field with the square-root of the photospheric pressure. This was derived from an analysis of a variety of theoretical spot models including those by Yun (1968). These were based on the Schliiter-Temesvary (S- T) similarity assumption. To answer criticisms that such modeling may have unphysical (non-axial maxima) solutions, the S-T model was revisited, Moon et al. (1998), with an improved vector potential function. We consider here the consequences of this work for the scaling relation. We show that by dimensionalizing the lateral force balance equation for the S- T model one finds that a single parameter enters as a characteristic value of the solution. This parameter yields Skumanich's scaling directly. Using an observed universal flux-radius relation for dark solar magnetic features (spots and pores) for comparison, we find good to fair agreement with Yun's characteristic value, however the Moon et al. values deviate significantly.

• Speech Sciences
• /
• v.9 no.1
• /
• pp.39-47
• /
• 2002

This paper proposes an effective method to remedy the acoustic modeling problem inherent in the usual log-normal Parallel Model Composition intended for achieving robust speech recognition. In particular, the Gaussian kernels under the prescribed log-normal PMC cannot sufficiently express the corrupted speech distributions. The proposed scheme corrects this deficiency by judiciously selecting the 'fairly' corrupted component and by re-estimating it as a mixture of two distributions using data-driven PMC. As a result, some components become merged while equal number of components split. The determination for splitting or merging is achieved by means of measuring the similarity of the corrupted speech model to those of the clean model and the noise model. The experimental results indicate that the suggested algorithm is effective in representing the corrupted speech distributions and attains consistent improvement over various SNR and noise cases.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.16 no.2
• /
• pp.119-124
• /
• 2016

According to the expansion of smartphone penetration and development of wearable device, personal context information can be easily collected. To use this information, the context aware recommender system has been actively studied. The key issue in this field is how to deal with the context information, as users are influenced by different contexts while rating items. But measuring the similarity among contexts is not a trivial task. To solve this problem, we propose context aware post-filtering to apply the context compensation. To be specific, we calculate the compensation for different context information by measuring their average. After reflecting the compensation of the rating data, the mechanism recommends the items to the user. Based on the item recommendation list, we recover the rating score considering the context information. To verify the effectiveness of the proposed method, we use the real movie rating dataset. Experimental evaluation shows that our proposed method outperforms several state-of-the-art approaches.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2003.11a
• /
• pp.121-124
• /
• 2003

It is a major concern of e-shopping mall managers to satisfy a variety of customer's desire by recommending a proper commodity to the expected purchaser. Customer information like customer's fondness and idiosyncrasy in shopping has not been used effectively for the customers or the suppliers. Conventionally, e-shopping mall managers have recommended specific items of commodities to their customers without considering thoroughly in a customer point of view. This study introduces the ways of a choosing and recommending of commodities for customer themselves or others. A similarity measure between one member's idiosyncrasy and the other members' is developed based on the rule base and the case base. The case base is improved by recognizing and learning the changes of customer's desire and shopping trend.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.15 no.11
• /
• pp.6810-6814
• /
• 2014

The concept of IFS (Iterated Function System) was applied to compress and transmit image data efficiently. To compress the image data with IFS, self-similarity was used to search a similar block. To improve the coding performance for the iterated function system with natural images, the image will be formed of properly transformed parts of itself to minimize the coding error. The simulation results using the proposed IFS represent high PSNR performance and improved compression efficiency with the coefficient of a recursive function.

• Proceedings of the Korea Contents Association Conference
• /
• 2009.05a
• /
• pp.663-667
• /
• 2009

본 논문에서는 인접블록간의 움직임 유사도를 이용하여 불필요한 후보블록을 보다 빠르게 제거하는 PDE기반의 고속 블록매칭 알고리즘을 제안한다. 제안한 방법은 기존의 방법보다 불필요한 계수를 효율적으로 제거하기 위하여 인접 블록간의 영상의 유사성에 기초하여 인접한 네개의 매크로블록 가운데 최대 복잡도를 가지는 서브블록의 누적된 비율(cumulative distribution function-CDF)을 사용하고 서브블록별 복잡도가 집중되지 않도록 하기위하여 normalized 기반 매칭스캔 방법을 사용하여 효율적으로 계산량을 줄였다. 제안한 알고리즘은 화질의 저하 없이 기존의 PDE 알고리즘에 비해 60% 이상의 계산량을 줄였으며, MPEG-2 및 MPEG-4 AVC를 이용하는 비디오 압축 응용분야에 유용하게 사용될 수 있을 것이다.