• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.05a
• /
• pp.487-490
• /
• 2021

최근 반려동물 관련 산업이 증가함에 따라 반려동물의 행동을 분석하는 연구가 진행되고 있다. 이를 바탕으로 본 논문에서는 반려동물 행동 분석을 통한 이상행동 예측 시스템을 제안한다. 이 시스템은 CCTV로부터 반려동물의 영상 데이터를 수집하고, YOLOv4(You Only Look Once version 4)를 통해 반려동물의 객체를 탐지한다. 행동을 분석하기 위해 탐지된 반려동물 객체를 DeepLabCut 딥러닝 알고리즘을 사용하여 관절 좌표 정보를 추출한다. 추출된 관절 좌표와 반려동물의 일반적인 행동을 매칭하여 이상행동을 예측하기 위한 DNN(Deep Neural Networks)의 입력 데이터로써 사용된다. 위 과정을 통해 반려동물의 전체적인 행동을 분석하여 이상행동을 예측한다. 이 시스템을 통해 반려동물의 행동을 분석하고 이상행동을 예측함으로써 반려동물 의료 관련 사업에도 적용될 수 있을 것이다.

• Architectural research
• /
• v.22 no.2
• /
• pp.41-52
• /
• 2020

Laboratory buildings with specialized equipment and ventilation systems pose challenges in terms of efficient energy use and initial construction costs. Additionally, lab spaces should have flexible and efficient layouts and provide a comfortable indoor research environment. Therefore, this study aims to identify the correlation between the facade of a building and its interior layout from case studies of energy-efficient research labs and to propose passive energy design strategies for the establishment of an optimal research environment. The case studies in this paper were selected from the American Institute of Architects Committee on the Environment Top Ten Projects and Leadership in Energy and Environmental Design (LEED) certified research lab projects. In this paper, the passive design strategies of space zoning, façade design devices to control heating and cooling loads were analyzed. Additionally, the relationships between these strategies and the interior lab layouts, lab support spaces, offices, and circulation areas were examined. The following four conclusions were drawn from the analysis of various cases: 1) space zoning for grouping areas with similar energy requirements is performed to concentrate similar heating and cooling demands to simplify the HVAC loads. 2) Public areas such as corridor, atrium, or courtyard can serve as buffer zones that employ passive solar design to minimize the mechanical energy load. 3) A balanced window-to-wall ratio (WWR), exterior shading devices, and natural ventilation systems are applied according to the space programming energy requirements to minimize the dependence on mechanical service. 4) Lastly, typical laboratory space zoning categories can be revised, reversed, and even reconfigured to minimize the energy load and adjust to the site context. This study can provide deep insights into various design strategies employed for construction of green laboratories along with intuitive arrangement of various building components such as laboratory spaces, lab support spaces, office spaces, and common public areas. The key findings of this study can contribute towards creating improved designs of laboratory facilities with reduced carbon footprint and greenhouse emissions.

• International Journal of Naval Architecture and Ocean Engineering
• /
• v.6 no.3
• /
• pp.749-761
• /
• 2014

Tubing adapter is a key connection device in Gas-Tight Deep-Sea Water Sampler (GTWS). The sealing performance of the tubing adapter directly affects the GTWS's overall gas tightness. Tubing adapters with good sealing performance can ensure the transmission of seawater samples without gas leakage and can be repeatedly used. However, the sealing performance of tubing adapters made of different materials was not studied sufficiently. With the research discussed in this paper, the materials match schemes of the tubing adapters were proposed. With non-linear finite element contact analysis and sea trials in the South China Sea, it is expected that the recommended materials match schemes not only meet the requirements of tubing adapters' sealing performance but also provide the feasible options for the following research on tubing adapters in GTWS.

• International Journal of Computer Science & Network Security
• /
• v.22 no.10
• /
• pp.406-412
• /
• 2022

Scene classification of very high-resolution (VHR) imagery can attribute semantics to land cover in a variety of domains. Real-world application requirements have not been addressed by conventional techniques for remote sensing image classification. Recent research has demonstrated that deep convolutional neural networks (CNNs) are effective at extracting features due to their strong feature extraction capabilities. In order to improve classification performance, these approaches rely primarily on semantic information. Since the abstract and global semantic information makes it difficult for the network to correctly classify scene images with similar structures and high interclass similarity, it achieves a low classification accuracy. We propose a VHR remote sensing image classification model that uses extracts the global feature from the original VHR image using an EfficientNet-V2L CNN pre-trained to detect similar classes. The image is then classified using a multilayer perceptron (MLP). This method was evaluated using two benchmark remote sensing datasets: the 21-class UC Merced, and the 38-class PatternNet. As compared to other state-of-the-art models, the proposed model significantly improves performance.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2015.08a
• /
• pp.84.1-84.1
• /
• 2015

Recent years in particular in Korea see intensive interests in a deep geothermal engineering and its application in different uses as far as from direct uses to power generation sectors, that are achieved by harnessing hot energy sources from the earth. For instance widespread interest has been generated because the geothermal energy is the source that one extracts it for more than 20 hours per day and for about 30 years of an operation of the plant, which enables to give base load as for heating as well as an electric generation. In retrospect, shallow geothermal energy using heat pumps is commonplace in Korea while the deep geothermal is in the early stage of the development. Geothermal energies in view of the way of extracting heat are mainly categorized into several types such as a single well system, a hydrothermal system, an enhanced geothermal system (EGS) etc. In this talk, this speaker focuses on the thermo-fluid engineering of the single well system by introducing the modeling in order to harness hot fluid that is thermally balanced with the fluid of an injection well, which provides a challenge to assess the life time of the well. To avoid the loss of the temperature in producing the hot fluid, a specialized pipe or a borehole heat exchanger has been designed, and its concept is introduced. On the other hand, a binary system or an organic Rankine cycle, which provides the methodology to convert the heat into an electricity, is briefly introduced. Some experimental results of the binary system which has been constructed in our lab will be presented. Lastly as for the future direction, some comments for the industrialization of the deep geothermal energy in this country will be discussed.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.32 no.4
• /
• pp.265-272
• /
• 2019

In this paper, a vision-based deep learning algorithm and image processing method are proposed to detect bolt-loosening in steel connections. To achieve this objective, the following approaches are implemented. First, a bolt-loosening detection method that includes regional convolutional neural network(RCNN)-based deep learning algorithm and Hough line transform(HLT)-based image processing algorithm are designed. The RCNN-based deep learning algorithm is developed to identify and crop bolts in a connection image. The HLT-based image processing algorithm is designed to estimate the bolt angles from the cropped bolt images. Then, the proposed vision-based method is evaluated for verifying bolt-loosening detection in a lab-scale girder connection. The accuracy of the RCNN-based bolt detector and HLT-based bolt angle estimator are examined with respect to various perspective distortions.

• Journal of radiological science and technology
• /
• v.45 no.2
• /
• pp.151-158
• /
• 2022

The Cardiac Gated Blood Pool (GBP) scintigram, a nuclear medicine imaging, calculates the left ventricular Ejection Fraction (EF) by segmenting the left ventricle from the heart. However, in order to accurately segment the substructure of the heart, specialized knowledge of cardiac anatomy is required, and depending on the expert's processing, there may be a problem in which the left ventricular EF is calculated differently. In this study, using the DeepLabV3 architecture, GBP images were trained on 93 training data with a ResNet-50 backbone. Afterwards, the trained model was applied to 23 separate test sets of GBP to evaluate the reproducibility of the region of interest and left ventricular EF. Pixel accuracy, dice coefficient, and IoU for the region of interest were 99.32±0.20, 94.65±1.45, 89.89±2.62(%) at the diastolic phase, and 99.26±0.34, 90.16±4.19, and 82.33±6.69(%) at the systolic phase, respectively. Left ventricular EF was calculated to be an average of 60.37±7.32% in the ROI set by humans and 58.68±7.22% in the ROI set by the deep learning segmentation model. (p<0.05) The automated segmentation method using deep learning presented in this study similarly predicts the average human-set ROI and left ventricular EF when a random GBP image is an input. If the automatic segmentation method is developed and applied to the functional examination method that needs to set ROI in the field of cardiac scintigram in nuclear medicine in the future, it is expected to greatly contribute to improving the efficiency and accuracy of processing and analysis by nuclear medicine specialists.

• Korean Journal of Remote Sensing
• /
• v.37 no.5_1
• /
• pp.1149-1161
• /
• 2021

With a trend of the utilization of computer vision for satellite images, cloud detection using deep learning also attracts attention recently. In this study, we conducted a U-Net cloud detection modeling using SPARCS (Spatial Procedures for Automated Removal of Cloud and Shadow) Cloud Dataset with the image data augmentation and carried out 10-fold cross-validation for an objective assessment of the model. Asthe result of the blind test for 1800 datasets with 512 by 512 pixels, relatively high performance with the accuracy of 0.821, the precision of 0.847, the recall of 0.821, the F1-score of 0.831, and the IoU (Intersection over Union) of 0.723. Although 14.5% of actual cloud shadows were misclassified as land, and 19.7% of actual clouds were misidentified as land, this can be overcome by increasing the quality and quantity of label datasets. Moreover, a state-of-the-art DeepLab V3+ model and the NAS (Neural Architecture Search) optimization technique can help the cloud detection for CAS500 (Compact Advanced Satellite 500) in South Korea.

• Korean Journal of Remote Sensing
• /
• v.40 no.1
• /
• pp.71-80
• /
• 2024

In this study, a proposed ensemble learning technique aims to enhance the semantic segmentation performance of images captured by Unmanned Aerial Vehicles (UAVs). With the increasing use of UAVs in fields such as urban planning, there has been active development of techniques utilizing deep learning segmentation methods for land cover segmentation. The study suggests a method that utilizes prominent segmentation models, namely U-Net, DeepLabV3, and Fully Convolutional Network (FCN), to improve segmentation prediction performance. The proposed approach integrates training loss, validation accuracy, and class score of the three segmentation models to enhance overall prediction performance. The method was applied and evaluated on a land cover segmentation problem involving seven classes: buildings,roads, parking lots, fields, trees, empty spaces, and areas with unspecified labels, using images captured by UAVs. The performance of the ensemble model was evaluated by mean Intersection over Union (mIoU), and the results of comparing the proposed ensemble model with the three existing segmentation methods showed that mIoU performance was improved. Consequently, the study confirms that the proposed technique can enhance the performance of semantic segmentation models.

• Journal of Internet Computing and Services
• /
• v.25 no.2
• /
• pp.101-111
• /
• 2024

In this paper, we propose a concpetual architecture and its implementation approach for contextualizing unstructured CCTV-video frame data into structured XML-video textual data by using the deep-learning neural network models and frameworks. Conclusively, through the conceptual architecture and the implementation approach proposed in this paper, we can eventually realize and implement the so-called sharable working and experiencing knowledge management platforms to be adopted to smart factories in various industries.