• Journal of Digital Convergence
• /
• v.19 no.6
• /
• pp.251-258
• /
• 2021

In this paper, we proposed an encoder-decoder model utilizing residual learning to improve the accuracy of the U-Net-based semantic segmentation method. U-Net is a deep learning-based semantic segmentation method and is mainly used in applications such as autonomous vehicles and medical image analysis. The conventional U-Net occurs loss in feature compression process due to the shallow structure of the encoder. The loss of features causes a lack of context information necessary for classifying objects and has a problem of reducing segmentation accuracy. To improve this, The proposed method efficiently extracted context information through an encoder using residual learning, which is effective in preventing feature loss and gradient vanishing problems in the conventional U-Net. Furthermore, we reduced down-sampling operations in the encoder to reduce the loss of spatial information included in the feature maps. The proposed method showed an improved segmentation result of about 12% compared to the conventional U-Net in the Cityscapes dataset experiment.

• Journal of Korean Society on Water Environment
• /
• v.37 no.3
• /
• pp.157-167
• /
• 2021

Korea needs to develop a rational system to separate stormwater utility fee from current sewerage fee. In this study, the scenario for calculating stormwater utility fee of Bupyeong-gu was suggested and the results were considered. For this purpose, the application of stormwater utility fee overseas and current domestic system were analyzed. A three step calculating scenario considering suitable domestic situation and impervious surface area was suggested. Water, sewerage usage, and hydrant data were collected. The total amount of water and sewerage fees for land use were calculated. The sewerage fee of Bupyeong-gu for the year 2014 was 21,685,446,578 won. Assuming that 40% of this amount was the cost associated to stormwater, the result showed that the fees for residential area in third step decreased by 0.77% compared to that of the first step. For commercial area, the stormwater utility fee decreased by 36.87%. For industrial area, although the consumption of water was similar to that of commercial area, the stormwater utility fee increased by 8.35%. For green area, the fee increased by 37.46%. This study demonstrated that the calculation of actual stormwater utility fee using impervious surface map and impervious Surface Ratio Estimation Methodology developed in previous studies is feasible.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.10
• /
• pp.1376-1383
• /
• 2020

Although studies on fire evacuation systems are increasing, studies on the evacuation of evacuees in indoor spaces are insufficient. According to the latest research, it has been suggested that the use of high frequency might be effective for identifying the location of evacuees indoors. Accordingly, in this paper, the authors intend to develop evacuation location recognition technology and fire evacuation guidance system using high-frequency and a smartphone. The entire system was developed, including an app server, evacuees location recognition unit, an evacuation route search, an output unit, and a speaker unit based on Wi-Fi communication. The experimental results proved the possibility of the effectiveness of the system in the fire situation data. It is expected that this study could be used as an essential study of a fire evacuation guidance system using high frequency data in case of fire.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.11
• /
• pp.491-500
• /
• 2021

In this study, we propose a method for detecting fall behavior using MS Kinect v2 RGBD Camera-based Human-Skeleton Keypoints and a 2-Stacked Bi-LSTM model. In previous studies, skeletal information was extracted from RGB images using a deep learning model such as OpenPose, and then recognition was performed using a recurrent neural network model such as LSTM and GRU. The proposed method receives skeletal information directly from the camera, extracts 2 time-series features of acceleration and distance, and then recognizes the fall behavior using the 2-Stacked Bi-LSTM model. The central joint was obtained for the major skeletons such as the shoulder, spine, and pelvis, and the movement acceleration and distance from the floor were proposed as features of the central joint. The extracted features were compared with models such as Stacked LSTM and Bi-LSTM, and improved detection performance compared to existing studies such as GRU and LSTM was demonstrated through experiments.

• Journal of Industrial Convergence
• /
• v.20 no.6
• /
• pp.7-15
• /
• 2022

Recently, due to the development of AI and big data technologies following the advent of the era of the 4th Industrial Revolution, the emerging metaverse industry is emerging as a new business, and in particular, from this point of view, this paper analyzes the history of metaverse and the pros and cons of "Geppetto", which is the most popular in the Korean metaverse market, and aims to give an appropriate direction for future development based on this. In order to carry out this study, we first used SWOT analysis techniques as an initial enterprise analysis method to examine the strengths and weaknesses, opportunities and threat requirements, and derive the status of each factor. Based on the factors in each of the subsequent derivatives, we wanted to explore the TOWS development strategy and present significant implications based on this.

• The Korean Journal of Applied Statistics
• /
• v.35 no.5
• /
• pp.593-601
• /
• 2022

We call "fruad" the cases that are not frequently occurring but cause significant losses. Fraud detection is commonly encountered in various applications, including wafer production in the semiconductor industry. It is not trivial to directly extend the standard binary classification methods to the fraud detection context because the misclassification cost is much higher than the normal class. In this article, we propose the functional fraud detection support vector machine (F²DSVM) that extends the fraud detection support vector machine (FDSVM) to handle functional covariates. The proposed method seeks a classifier for a function predictor that achieves optimal performance while achieving the desired sensitivity level. F²DSVM, like the conventional SVM, has piece-wise linear solution paths, allowing us to develop an efficient algorithm to recover entire solution paths, resulting in significantly improved computational efficiency. Finally, we apply the proposed F²DSVM to the defective wafer detection problem and assess its potential applicability.

• Korean journal of applied entomology
• /
• v.61 no.2
• /
• pp.307-311
• /
• 2022

The genus Areotetes van Achterberg & Li, 2013 (Hymenoptera: Braconidae: Opiinae), which is endoparasitoid of mining or infesting of fruit dipterous larvae, have been reported for the first time in China. Currently, four species of the genus Areotetes have been known from the province Hunan and Fujian, China. In this study the genus Areotetes with Areotetes carinuliferus van Achterberg & Li, 2013 is reported for the first time from Korea. Material studied in the present study were collected by sweeping in Mt Gongchi, Eochungdo, Province Jeonbuk, Korea. Herein, diagnosis of genus, description, distribution, and diagnostic illustration of A. carinuliferus are provided. In addition, DNA barcode data of the partial gene of mitochondrial cytochorome c oxidase subunit I (COI) are included.

• Journal of the Korean Institute of Gas
• /
• v.26 no.4
• /
• pp.18-26
• /
• 2022

The Korean government is conducting heat-using equipment(industrial boiler) inspection in accordance with the Energy Use Rationalization Act because of the heat-using equipment(industrial boiler)'s risks such as explosion and fire, and safe use and management. This paper aimed to setup the safe and performance vulnerabilities from database based on the inspection results for heat-using equipment(industrial boiler). This study surveyed the inspection results of 1,249 heat-using equipment(industrial boiler) which were failed inspection of heat-using equipment(industrial boiler) from january 2016 to december 2020. And the analysis method is to inform safety and performance vulnerability categories of heat-using equipment(industrial boiler) by statistically analyzing the failure reasons of boiler type and inspection type which are high variance in failure rate. The safety and performance vulnerability categories was abbreviated into 18 cases. And each catagory's main reason for failure was suggested by additional analyzing the opinions of inspectors. This paper would be the basic source and the comprehensive information dealing with the safety and performance vulnerability of heat-using equipment(industrial boiler).

• The Journal of the Convergence on Culture Technology
• /
• v.8 no.5
• /
• pp.705-712
• /
• 2022

The pandemic over the past three years has brought drastic changes in our lives, and those changes are now becoming a natural part of our daily lives. Daily life and economic activities based on online, such as video conferencing, remote classes, telecommuting, and online streaming services, have become daily routines after Corona. And the result of rapid development of communication and graphic technology is the metaverse. Therefore, the purpose of this study is to study the possibility from the perspective of public design for the correct use of the metaverse space of local governments. To this end, in this study, big data analysis was performed on 'local government', 'public design', and 'metaverse'. As a result of this study, we should actively use metaverse with high topic and potential as a space for local governments' promotion or discussion as a means to restore the reliability of local governments and overcome negative perceptions. In addition, it is necessary to actively reflect public design in order to increase the public reliability of local governments' metaverse.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.28 no.7
• /
• pp.1274-1280
• /
• 2022

The resistance performance evaluation of general ships using computational fluid dynamics requires a lot of time and cost, and various methods are being studied to reduce the time and cost. Existing methods using main particulars or cross sections of ships have limitations in estimating resistance performance that is greatly dependent on the shape of the ship. In this paper, we propose a deep neural network model that can quickly predict the resistance performance of the hull surface by inputting the geometric information of the hullform mesh. The proposed deep neural network model based on Perceiver IO can immediately predict resistance performance, unlike computational fluid dynamics techniques that require calculation in each time step. It shows the result of estimating the resistance performance with an average error of less than 1% in the data set for a 50 K tanker ship, a type of low-speed full ship.