• 웰빙융합연구
• /
• 제5권2호
• /
• pp.15-19
• /
• 2022

Purpose: This pilot study aimed to clarify features of motor module during walking in exercise experts who experienced lately repeated training for sports skill. To identify motor modules, autoencoder machine learning algorithm was used, and modules were extracted from muscle activities of lower extremities. Research design, data and methodology: A total of 10 university students were participated. 5 students did not experience any sports training before, and 5 students did experience sports training more than 5 years. Eight muscle activities of dominant lower extremity were measured. After modules were extracted by autoencoder, the numbers of modules and spatial muscle weight values were compared between two groups. Results: There was no significant difference in the minimal number of motor modules that explain more than 90% of original data between groups. However, in similarity analysis, three motor modules were shown high similarity (r>0.8) while one module was shown low similarity (r<0.5). Conclusions: This study found not only common motor modules between exercise novice and expert during walking, but also found that a specific motor module, which would be associated with high motor control ability to distinguish the level of motor performance in the field of sports.

• 정보과학회 컴퓨팅의 실제 논문지
• /
• 제23권9호
• /
• pp.535-541
• /
• 2017

소프트웨어 시장 규모가 확대되면서 다양한 요구사항을 만족시키는 대규모 소프트웨어가 개발되고 있다. 이로 인해 소프트웨어 복잡도가 증가하게 되고 품질 관리는 어려워졌다. 특히, 레거시 시스템의 개선 및 새로운 시스템 개발 환경에서 재사용은 중요하다. 이 논문에서는 품질을 인증 받은 모듈을 재사용하는 방법을 제안한다. 재사용 적용 레벨은 코드 영역(메소드, 클래스, 컴포넌트), 프로젝트 도메인, 비즈니스로 나누어진다. 이 논문에서는 소프트웨어 복잡성을 측정하는 결합도와 응집도 기반의 재사용 메트릭과 메소드와 클래스 레벨에 따라 "재사용에 적합한 모듈 덩어리"를 시각화하는 재사용 적합 모듈 추출 메카니즘을 제안한다. 레거시 프로젝트에 역공학 기법을 적용하여 어떤 모듈/객체/덩어리가 재사용할 수 있는 지를 식별하고 확장 시스템을 개발하거나 유사한 새로운 시스템을 개발하기 위해 재사용한다면 소프트웨어의 신뢰성을 보장하고 소프트웨어 개발 단계에서 필요한 시간과 비용을 절감시킬 수 있다.

• 전기전자학회논문지
• /
• 제11권4호
• /
• pp.257-264
• /
• 2007

MM22 마이크로트론은 1986년도 11월부터 2006년 2월까지 암 환자를 위한 방사선 치료 장비로 사용되었다. 장비의 노후로 치료 장비에서 연구 및 교육용으로 전환하기 위해 방사선의학연구센터로 이전 설치하였다. 본 논문에서는 이전 설치 한 후 빔 인출을 수행하기 위해 마이크로트론의 동작원리, 시스템을 구성하는 각 장치의 특성을 분석하여 보았고 주요 부분의 파라메타인 펄스구조의 특징을 살펴보았다. 실제, 각 주요 시스템의 펄스를 측정하였고 빔 인출부, 빔 라인 및 최종단인 타겟에서 빔 인출 기법을 통하여 빔 인출 및 빔 측정을 수행하였다. 이전 설치 후 10 MV X-선의 경우 최종 단 치료기에서 30 mA 타겟 전류를 인출하였고, 필름을 SSD 100 cm, $10{\times}10cm^2$ 조사면에 놓고 100 MU 방사선을 조사하였다. 조사면의 방사선분포의 평탄도 측정 결과 3%이내로 안정적인 빔을 인출하여 이전설치가 성공적으로 수행되었음을 확인할 수 있었다.

• Journal of international Conference on Electrical Machines and Systems
• /
• 제2권4호
• /
• pp.441-446
• /
• 2013

A thermal model of power modules based on the physical dimension and thermal properties is proposed in this paper. The heat path in the power module is considered as a one-dimensional heat transfer in the model. The method of the parameters extraction for the model is given in the paper. With high speed and accuracy, the thermal model is suit for electrothermal simulation. The proposed model is verified by experimental results.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권7호
• /
• pp.2407-2424
• /
• 2022

Recently, Transformer has made great progress in video retrieval tasks due to its high representation capability. For the structure of a Transformer, the cascaded self-attention modules are capable of capturing long-distance feature dependencies. However, the local feature details are likely to have deteriorated. In addition, increasing the depth of the structure is likely to produce learning bias in the learned features. In this paper, an improved Transformer structure named TransDCS (Transformer with Dynamic Convolution and Shortcut) is proposed. A Multi-head Conv-Self-Attention module is introduced to model the local dependencies and improve the efficiency of local features extraction. Meanwhile, the augmented shortcuts module based on a dual identity matrix is applied to enhance the conduction of input features, and mitigate the learning bias. The proposed model is tested on MSRVTT, LSMDC and Activity-Net benchmarks, and it surpasses all previous solutions for the video-text retrieval task. For example, on the LSMDC benchmark, a gain of about 2.3% MdR and 6.1% MnR is obtained over recently proposed multimodal-based methods.

• 인터넷정보학회논문지
• /
• 제21권5호
• /
• pp.21-29
• /
• 2020

Accurate estimation of human pose relies on backbone method in which its role is to extract feature map. Up to dated, the method of backbone feature extraction is conducted by the plain convolutional neural networks named by CNN and the residual neural networks named by Resnet, both of which have various architectures and performances. The CNN family network such as VGG which is well-known as a multiple stacked hidden layers architecture of deep learning methods, is base and simple while Resnet which is a bottleneck layers architecture yields fewer parameters and outperform. They have achieved inspired results as a backbone network in human pose estimation. However, they were used then followed by different pose estimation networks named by pose parsing module. Therefore, in this paper, we present a comparison between the plain CNN family network (VGG) and bottleneck network (Resnet) as a backbone method in the same pose parsing module. We investigate their performances such as number of parameters, loss score, precision and recall. We experiment them in the bottom-up method of human pose estimation system by adapted the pose parsing module of openpose. Our experimental results show that the backbone method using VGG network outperforms the Resent network with fewer parameter, lower loss score and higher accuracy of precision and recall.

• International Journal of Internet, Broadcasting and Communication
• /
• 제10권2호
• /
• pp.45-50
• /
• 2018

Recently the water management department advised the water treatment industry to focus on deploy the chemical free and the environmentally responsible process to adopt on water treatment plants in every country. In this objective, water treatment process started using ultrasonic based phycocyanin extraction with fluorescence measurement techniques to detect the change in the yield of phycocyanin. This paper propose the design of optical filter model for fluorescence technique based immersive optical phycocyanin measurement sensor design. The proposed design uses the multi-wavelength sensor module for irradiating part, and this plays a role of removing a wavelength band other than 590 ~ 620 nm. The preliminary study on immersed phycocyanin sensor, the fluorescence value of picocyanin according to the ultrasonic intensity, treatment time and number of cells was measured using JM phycocyanin module to emulate the proposed design, and were compared performance of the proposed sensor emulation. In this design, the phycocyanin fluorescence value increased about 2.1 ~ 4.7 times as the ultrasonic treatment time increased as compared with JM phycocyanin module, and the phycocyanin fluorescence value within the analysis range was obtained by ultrasonic treatment within one minute.

• 한국멀티미디어학회논문지
• /
• 제23권12호
• /
• pp.1540-1551
• /
• 2020

Aiming at the problem that the existing human behavior recognition algorithm cannot fully utilize the multi-level spatio-temporal information of the network, a human behavior recognition algorithm based on a dense three-dimensional residual network is proposed. First, the proposed algorithm uses a dense block of three-dimensional residuals as the basic module of the network. The module extracts the hierarchical features of human behavior through densely connected convolutional layers; Secondly, the local feature aggregation adaptive method is used to learn the local dense features of human behavior; Then, the residual connection module is applied to promote the flow of feature information and reduced the difficulty of training; Finally, the multi-layer local feature extraction of the network is realized by cascading multiple three-dimensional residual dense blocks, and use the global feature aggregation adaptive method to learn the features of all network layers to realize human behavior recognition. A large number of experimental results on benchmark datasets KTH show that the recognition rate (top-l accuracy) of the proposed algorithm reaches 93.52%. Compared with the three-dimensional convolutional neural network (C3D) algorithm, it has improved by 3.93 percentage points. The proposed algorithm framework has good robustness and transfer learning ability, and can effectively handle a variety of video behavior recognition tasks.

• 국제학술발표논문집
• /
• The 9th International Conference on Construction Engineering and Project Management
• /
• pp.1202-1208
• /
• 2022

Techniques based on computer vision are becoming increasingly important in construction safety monitoring. Using AI algorithms can automatically identify conceivable hazards and give feedback to stakeholders. However, the construction site remains various potential hazard situations during the project. Due to the site complexity, many visual devices simultaneously participate in the monitoring process. Therefore, it challenges developing and operating corresponding AI detection algorithms. Safety information resulting from computer vision needs to organize before delivering it to safety managers. This study proposes a framework for computer vision-aided construction safety monitoring using collaborative 4D BIM information to address this issue, called CSM4D. The suggested framework consists of two-module: (1) collaborative BIM information extraction module (CBIE) extracts the spatial-temporal information and potential hazard scenario of a specific activity; through that, Computer Vision-aid Safety Monitoring Module (CVSM) can apply accurate algorithms at the right workplace during the project. The proposed framework is expected to aid safety monitoring using computer vision and 4D BIM.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2023년도 춘계학술발표대회
• /
• pp.510-512
• /
• 2023

In order to enhance a model's capability for detecting facial expressions, this research suggests a pipeline that makes use of the GradCAM component. The patching module and the pseudo-labeling module make up the pipeline. The patching component takes the original face image and divides it into four equal parts. These parts are then each input into a 2Dconvolutional layer to produce a feature vector. Each picture segment is assigned a weight token using GradCAM in the pseudo-labeling module, and this token is then merged with the feature vector using principal component analysis. A convolutional neural network based on transfer learning technique is then utilized to extract the deep features. This technique applied on a public dataset MMI and achieved a validation accuracy of 96.06% which is showing the effectiveness of our method.