• 디지털융복합연구
• /
• 제17권10호
• /
• pp.259-264
• /
• 2019

본 논문에서는 엔터프라이즈 환경에서의 딥 러닝에 대한 이미지 예측 시스템 아키텍처를 제안한다. 엔터프라이즈 환경에 대해 인공지능 플랫폼으로 변환을 쉽게 하고, 인공지능 플랫폼이 파이선에 집중되어서 자바 중심의 엔터프라이즈 개발이 어려운 단점을 개선하기 위해 자바 중심의 아키텍처에서도 충분한 딥 러닝 서비스의 개발과 수정이 가능하도록 한다. 또한, 제안된 환경을 토대로 이미지 예측 실험을 통해 기존에 학습된 딥 러닝 아키텍처 환경에서의 정확도가 높은 예측 시스템을 제안한다. 실험을 통해 딥 러닝이 수행되기 위해 제공된 이미지 예에서 95.23%의 정확도를 보이며, 제안된 모델은 유사한 다른 모델에 비교해 96.54%의 정확도를 보인다. 제시된 아키텍처를 활용하여 활발한 엔터프라이즈급 환경의 딥 러닝 서비스가 개발 및 제공될 것으로 보이며, 기존 엔터프라이즈 환경이 딥 러닝 아키텍처가 탑재된 환경으로 전환이 활발히 이루어질 것이다.

• 한국정보통신학회논문지
• /
• 제26권6호
• /
• pp.859-864
• /
• 2022

매년 교통사고의 가장 큰 원인으로 손꼽히는 졸음운전은 운전자의 수면 부족, 산소 부족, 긴장감의 저하, 신체의 피로 등과 같은 다양한 요인을 동반한다. 졸음 유무를 확인하는 일반적인 방법으로 운전자의 표정과 주행패턴을 파악하는 방법, 심전도, 산소포화도, 뇌파와 같은 생체신호를 분석하는 방법들이 연구되고 있다. 본 논문은 영상을 검출하는 딥러닝 모델과 생체 신호 측정 기술을 이용한 운전자 피로 감지 시스템을 제안한다. 제안 방법은 일차적으로 딥러닝을 이용하여 운전자의 눈 모양과 하품 유무, 졸음으로 예상되는 신체 동작을 파악하여 졸음 상태를 감지한다. 이차적으로 맥파 신호와 체온을 이용하여 운전자의 피로 상태를 파악하여 시스템의 정확도를 높이도록 설계하였다. 실험 결과, 실시간 영상에서 운전자의 졸음 유무 판별이 안정적으로 가능하였으며 각성상태와 졸음 상태에서의 분당 심박수와 체온을 비교하여 본 연구의 타당성을 확인할 수 있었다.

• 문화기술의 융합
• /
• 제9권5호
• /
• pp.721-727
• /
• 2023

국가 운영에서 부정, 낭비, 남용을 방지하기 위해 감사 및 감찰기구를 운영하고, 이들에게 업무의 독립성 보장을 포함한 많은 권한과 책임을 부여하고 있다. 하지만 이들 기관에서 발생하는 문제나 비위에 대해서는 누가 어떻게 감독하는지에 대한 문제가 제기된다. 이러한 감독을 위해 미국에서는 감찰관 법을 통해 감사기구 간 협의체인 감사공동체를 구성하여 업무 수행의 기준 준수 여부와 감사기구 직원의 비위에 대한 조사를 수행하고, 의회, 대통령, 회계감사원, 소속 기관 지도부 등 다양한 이해관계자가 상호작용을 하면서 독립성 보장과 협업 문화를 조성하고 있다. 정부 운영의 효율성과 효과성을 제고한다는 측면에서 감독기구와 의회의 관계를 발전시킬 필요가 있는 것이다. 이에 본 논문에서는 이러한 미국 사례를 연구하여 우리나라 감사기구에 적용할 감독제도의 효율적인 정책방안을 제시한다.

• Safety and Health at Work
• /
• 제14권4호
• /
• pp.368-374
• /
• 2023

Background: Recognizing that access to safe and healthy working conditions is a human right, the World Health Organization (WHO) calls for specific occupational safety and health (OSH) programs for health workers (HWs). The WHO health systems' building blocks, and the International Labour Organization (ILO), highlight the importance of information as part of effective systems. This study examined how OSH stakeholders access, use, and value an occupational health information system (OHIS). Methods: A cross-sectional survey of OSH stakeholders was conducted as part of a larger quasi experimental study in four teaching hospitals. The study hospitals and participants were purposefully selected and data collected using a modified questionnaire with both closed and open-ended questions. Quantitative analysis was conducted and themes identified for qualitative analysis. Ethics approval was provided by the University of Pretoria and University of British Columbia. Results: There were 71 participants comprised of hospital managers, health and safety representatives, trade unions representatives and OSH professionals. At least 42% reported poor accessibility and poor timeliness of OHIS for decision-making. Only 50% had access to computers and 27% reported poor computer skills. When existing, OHIS was poorly organized and needed upgrades, with 85% reporting the need for significant reforms. Only 45% reported use of OHIS for decision-making in their OSH role. Conclusion: Given the gap in access and utilization of information needed to protect worker's rights to a safe and healthy workplace, more attention is warranted to OHIS development and use as well as education and training in South Africa and beyond.

• International journal of advanced smart convergence
• /
• 제8권4호
• /
• pp.104-112
• /
• 2019

This study proposed a robust detection algorithm. It detects face more stably with respect to changes in light and rotation forthe identification of a face shape. The proposed algorithm uses face shape asinput information in a single camera environment and divides only face area through preprocessing process. However, it is not easy to accurately recognize the face area that is sensitive to lighting changes and has a large degree of freedom, and the error range is large. In this paper, we separated the background and face area using the brightness difference of the two images to increase the recognition rate. The brightness difference between the two images means the difference between the images taken under the bright light and the images taken under the dark light. After separating only the face region, the face shape is recognized by using the self-organization feature map (SOFM) algorithm. SOFM first selects the first top neuron through the learning process. Second, the highest neuron is renewed by competing again between the highest neuron and neighboring neurons through the competition process. Third, the final top neuron is selected by repeating the learning process and the competition process. In addition, the competition will go through a three-step learning process to ensure that the top neurons are updated well among neurons. By using these SOFM neural network algorithms, we intend to implement a stable and robust real-time face shape recognition system in face shape recognition.

• 한국멀티미디어학회논문지
• /
• 제18권8호
• /
• pp.960-967
• /
• 2015

In this paper, we propose a user interface technique based on embedded sensors applying to apps in smart phone. Especially, we implement avata generation application using image filtering technique for photo image in smart phone. In the application, The embedded sensors are used as intuitive user interface to adjust the image filtering effect for making user satisfied effect in real time after the system produced the image filtering effect for avatar. This technique provides not a simple typed method of parameter values adjustment but a new intuitively emotional adjustment method in image filtering applications. The proposed technique can use sound values from embedded mike sensor for adjusting key values of sketch filter effect if the smart phone user produces sound. Similiarly the proposed technique can use coordinate values from embedded acceleration sensor for adjusting masking values of oil painting filter effect and use brightness values from embedded light sensor for adjusting masking values of sharp filter effect. Finally, we implement image filtering application and evaluate efficiency and effectiveness for the proposed technique.

• 대한수학회지
• /
• 제41권5호
• /
• pp.773-793
• /
• 2004

We consider a system of particles with locations { $X_{i}$ $^{n}$ (t):t$\geq$0,i＝1,…,n} in $R^{d}$ , time-varying weights { $A_{i}$ $^{n}$ (t) : t $\geq$0,i ＝ 1,…,n} and weighted empirical measure processes $V^{n}$ (t)＝1/n$\Sigma$$_{i=1}$$^{n}$ $A_{i}$ $^{n}$ (t)$\delta$ $X_{i}$ $^{n}$ (t), where $\delta$$_{x}$ is the Dirac measure. It is known that there exists the limit of { $V_{n}$ } in the week* topology on M( $R^{d}$ ) under suitable conditions. If { $X_{i}$ $^{n}$ , $A_{i}$ $^{n}$ , $V^{n}$ } satisfies some diffusion equations, applying Ito formula, we prove a central limit type theorem for the empirical process { $V^{n}$ }, i.e., we consider the convergence of the processes η$_{t}$ $^{n}$ ≡ n( $V^{n}$ -V). Besides, we study a characterization of its limit.t.

• Journal of Electrical Engineering and Technology
• /
• 제10권6호
• /
• pp.2384-2392
• /
• 2015

The technical development and practical applications of big-data for health is one hot topic under the banner of big-data. Big-data medical image fusion is one of key problems. A new fusion approach with coding based on Spherical Coordinate Domain (SCD) in Wireless Sensor Network (WSN) for big-data medical image is proposed in this paper. In this approach, the three high-frequency coefficients in wavelet domain of medical image are pre-processed. This pre-processing strategy can reduce the redundant ratio of big-data medical image. Firstly, the high-frequency coefficients are transformed to the spherical coordinate domain to reduce the correlation in the same scale. Then, a multi-scale model product (MSMP) is used to control the shrinkage function so as to make the small wavelet coefficients and some noise removed. The high-frequency parts in spherical coordinate domain are coded by improved SPIHT algorithm. Finally, based on the multi-scale edge of medical image, it can be fused and reconstructed. Experimental results indicate the novel approach is effective and very useful for transmission of big-data medical image(especially, in the wireless environment).

• International Journal of Contents
• /
• 제12권4호
• /
• pp.62-68
• /
• 2016

This paper describes an attempt to integrate flip teaching into a language classroom by adopting wikispaces as an online learning platform. The purpose of this study is to examine student perceptions of the effectiveness of using video lectures and wikispaces to foster active participation and collaborative learning. Flipped learning was implemented in an English writing class over one semester. Participants were 27 low intermediate level Korean university students. Data collection methods included background questionnaires at the beginning of the semester, learning experience questionnaires at the end of the semester, and semi-structured interviews with 6 focal participants. Because of the significance of video lectures in flip teaching, oCam was used for making weekly online lectures as a way of pre-class activities. Every week, online lectures were posted on the school LMS system (moodle). Every week, participants met in a computer room to perform in-class activities. Both in-class activities and post-class activities were managed by wikispaces. The results indicate that the flipped classroom facilitated student learning in the writing class. More than 53% of the respondents felt that it was useful to develop writing skills in a flipped classroom. Particularly, students felt that the video lectures prior to the class helped them improve their grammar skills. However, with respect to their satisfaction with collaborative works, about 44% of the participants responded positively. Similarly, 44% of the participants felt that in-class group work helped them interact with the other group members. Considering these results, this paper concludes with pedagogical suggestions and implications for further research.

• 한국항행학회논문지
• /
• 제12권2호
• /
• pp.164-169
• /
• 2008

본 논문에서는 영상분할에 의한 동작 추적 기법을 이용하여 침입자를 감시하고 관련 정보를 모바일 기반으로 확인하도록 하였다. 먼저, 탐지하고자 하는 일정한 영역을 촬영한 동영상에서 프레임을 추출하고, 인접한 두 프레임 사이의 이미지 차를 사용하여, 고정된 배경과 움직이는 대상을 분할한다. 분할된 전경 물체에서 에지를 검출하여 지정된 위치별로 추출된 에지의 중간 값을 추정하여 동작을 분석함으로써 침입자를 감시할 수 있도록 하였다. 동작이 검출되면, 영상은 WAP 풀 기반 영상 전송 방법을 사용하여 모바일 클라이언트로 전송한다.