• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.4
• /
• pp.189-198
• /
• 2023

As crimes frequently occur on the street, the spread of CCTV is increasing. However, due to the shortcomings of passively operated CCTV, the need for intelligent CCTV is attracting attention. Due to the heavy system of such intelligent CCTV, high-performance devices are required, which has a problem in that it is expensive to replace the general CCTV. To solve this problem, an intelligent CCTV system that recognizes low-quality images and operates even on devices with low performance is required. Therefore, this paper proposes a Saying CCTV system that can detect threats in real time by using the AWS cloud platform to lighten the system and convert images into text. Based on the data extracted using YOLO v4 and OpenPose, it is implemented to determine the risk object, threat behavior, and threat situation, and calculate the risk using machine learning. Through this, the system can be operated anytime and anywhere as long as the network is connected, and the system can be used even with devices with minimal performance for video shooting and image upload. Furthermore, it is possible to quickly prevent crime by automating meaningful statistics on crime by analyzing the video and using the data stored as text.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.29 no.4
• /
• pp.372-379
• /
• 2023

International oil prices are expected to increase from $85 a barrel this year to up to $100 a barrel in the second half of the year; this is likely to increase orders for offshore plants in the global market. One main characteristic of offshore plants is that a large helideck is located on the top side, and aluminum alloys are used as the basic material of the structure for weight reduction and corrosion resistance. Shipowners are increasing the size of helicopters to quickly evacuate lives in the event of an emergency, and the safety use load of devices that can stably secure helicopters to the deck is also required to increase. Owing to the nature of the aluminum material, the structural strength caused by welding is greatly reduced; therefore, the fixing device must be designed by embedding it in the deck and fixing it with bolts. In this study, a model applying aluminum alloy 6082-T6 was developed to develop a helicopter fastening device that can be used for large helidecks (diameter = 28 m). The developed item was verified through nonlinear structural strength calculation to satisfy the load used for the actual fastening condition. The load condition with a 45° showed a lower ultimate strength than the 90° case owing to local plastic collapse. The nonlinear structural collapse behavior showed a result similar to that of the experimental test. The main contents derived from this study are considered to be reference materials when evaluating the structural strength of similar aluminum equipment.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_4
• /
• pp.1165-1172
• /
• 2023

Recently, Korea has declared carbon neutrality by 2050. The Korea Forest Service is promoting the precision and high technology of forest resource surveys. As such, the demand for forest resource management is increasing, and the need to build a digital twin of forest space is increasing. However, to date, digital twin has only built and provided virtual city services, which are city and nationwide digital twin environments. Three-dimensional digital twin services targeting forest space are not operated and provided. Therefore, in this study, we aimed to implement a forest digital twin environment to provide 3D forest spatial information services corresponding to vertical information such as tree-level height and thorax diameter. By lightweighting realistic 3D tree models and applying 3D Tiles, we confirmed the feasibility of implementing a forest digital twin environment for 3D forest spatial information services. Through continuous research, we plan to implement a forest digital twin that can deploy and service 3D tree models for trees nationwide, including street trees in urban areas. This is expected to enable the development of forest digital twin services for forest resource management.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.89-95
• /
• 2024

Nowadays, the competitiveness of a company depends on the ability of all organizational members to share and utilize the organizational knowledge accumulated by the organization. As if to prove this, the world is now focusing on ChetGPT service using generative AI technology based on LLM (Large Language Model). However, it is still difficult to apply the ChetGPT service to work because there are many hallucinogenic problems. To solve this problem, sLLM (Lightweight Large Language Model) technology is being proposed as an alternative. In order to construct sLLM, corporate data is essential. Corporate data is the organization's ERP data and the company's office document knowledge data preserved by the organization. ERP Data can be used by directly connecting to sLLM, but office documents are stored in file format and must be converted to data format to be used by connecting to sLLM. In addition, there are too many technical limitations to utilize office documents stored in file format as organizational knowledge information. This study proposes a method of storing office documents in DB format rather than file format, allowing companies to utilize already accumulated office documents as an organizational knowledge system, and providing office documents in data form to the company's SLLM. We aim to contribute to improving corporate competitiveness by combining AI technology.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.1
• /
• pp.317-326
• /
• 2024

Biometric recognition is a technology that determines whether a person is identified by extracting information on a person's biometric and behavioral characteristics with a specific device. Cyber threats such as forgery, duplication, and hacking of biometric characteristics are increasing in the field of biometrics. In response, the security system is strengthened and complex, and it is becoming difficult for individuals to use. To this end, multiple biometric models are being studied. Existing studies have suggested feature fusion methods, but comparisons between feature fusion methods are insufficient. Therefore, in this paper, we compared and evaluated the fusion method of multiple biometric models using fingerprint, face, and iris images. VGG-16, ResNet-50, EfficientNet-B1, EfficientNet-B4, EfficientNet-B7, and Inception-v3 were used for feature extraction, and the fusion methods of 'Sensor-Level', 'Feature-Level', 'Score-Level', and 'Rank-Level' were compared and evaluated for feature fusion. As a result of the comparative evaluation, the EfficientNet-B7 model showed 98.51% accuracy and high stability in the 'Feature-Level' fusion method. However, because the EfficietnNet-B7 model is large in size, model lightweight studies are needed for biocharacteristic fusion.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.3
• /
• pp.1-12
• /
• 2024

In this paper, we fine-tuned KoAlpaca, a large-scale Korean language model, and implemented a blog text generation system utilizing it. Blogs on social media platforms are widely used as a marketing tool for businesses. We constructed training data of positive reviews through emotion analysis and refinement of collected sponsored blog texts and applied QLoRA for the lightweight training of KoAlpaca. QLoRA is a fine-tuning approach that significantly reduces the memory usage required for training, with experiments in an environment with a parameter size of 12.8B showing up to a 58.8% decrease in memory usage compared to LoRA. To evaluate the generative performance of the fine-tuned model, texts generated from 100 inputs not included in the training data produced on average more than twice the number of words compared to the pre-trained model, with texts of positive sentiment also appearing more than twice as often. In a survey conducted for qualitative evaluation of generative performance, responses indicated that the fine-tuned model's generated outputs were more relevant to the given topics on average 77.5% of the time. This demonstrates that the positive review generation language model for sponsored content in this paper can enhance the efficiency of time management for content creation and ensure consistent marketing effects. However, to reduce the generation of content that deviates from the category of positive reviews due to elements of the pre-trained model, we plan to proceed with fine-tuning using the augmentation of training data.

• Science of Emotion and Sensibility
• /
• v.13 no.3
• /
• pp.581-596
• /
• 2010

The purposes of this study are to investigate and analyze the level of information and the state of digital divide of the disabled by surveying the demand for information and communication assistive devices, and to provide basic data for plans on the development and utilization of information and communication assistive devices. In order to understand the actual condition and the state of digital divide of persons with disabilities, the differences of possession and accessibility of information technology devices, usage ability, and utilization were analyzed according to the disability profile by using the T-test. The results show that there are significant differences (T=-2.510*) of possession and accessibility of information technology devices with respect to the disability profile, and that the disabled have lower possession and accessibility of devices than the non-disabled. Result of this study's demand forecast shows that about 28% of total respondents are currently using information and communication assistive devices, and a majority (67%) of them answered that the use of assistive devices lend great help to their lives. The proportion of those who have been supported by the government or related organizations with information and communication assistive devices were 36% of the total respondents, and those satisfied with the performance of the devices were 46% of the total responses. Meanwhile, only 36% of total users answered that the operation and use of functions of the devices was easy and convenient, responding that the difficulty of operating assistive devices was the greatest inconvenience. Moreover, the general requests of respondents in regards to the devices were stabilization of device performance, miniaturization of size, simplification of buttons, and reduction of weight.

• Pediatric Gastroenterology, Hepatology & Nutrition
• /
• v.10 no.2
• /
• pp.138-146
• /
• 2007

Purpose: Ischemic enterocolitis (IEC) is the total or partial infarction of the intestine in the absence of occlusion of a major mesenteric blood vessel. The purpose of this study was to evaluate the clinical features of IEC in children. Methods: A clinical analysis of 6 patients with IEC who were admitted to the Department of Pediatrics at Pusan National University Hospital, between 1996 and 2005 was conducted retrospectively. Patients were diagnosed with IEC based on clinical characteristics, including radiologic, endoscopic, histopathologic, and intraoperative findings. Results: Four boys and 2 girls between the age of 6 weeks and 6 years were included in this study. Most of the patients were born at term and had a birth weight that was appropriate for their gestational age. The major symptoms of IEC observed included hematochezia or hematemesis (5 cases), vomiting, diarrhea, abdominal pain or irritability (4 cases), as well as abdominal distension and fever (3 cases). IEC occurred in thecolon in 5 cases (2 descending colon, 1 descending and sigmoid colon, 1 sigmoid colon, 1 whole colon) and the duodenal bulb and gastric antrum in 1 case each. The type of the lesions observed includedulcera, which were found in 3 cases, perforation, which was pbserved in 2 cases, necrotic patches, which were observed in 2 cases, stricture, which was observedin 1 cases, and massive membranous desquamation of the epithelium, which was observed in 1 case. Two of the patients received surgical treatment and the remaining four were treated conservatively. None of the patients died. Conclusion: The presentation of IEC varies, and the findings of this study will be helpful in managing patients with IEC.

• Journal of Bio-Environment Control
• /
• v.29 no.4
• /
• pp.440-447
• /
• 2020

The demand for large-scale horticultural complexes utilizing reclaimed lands is increasing, and one of the pending issues for the construction of large-scale facilities is to establish foundation design criteria. In this paper, we tried to review previous studies on the method of reinforcing the foundation of soft ground. Target construction methods are spiral piles, wood piles, crushed stone piles and PF (point foundation) method. In order to evaluate the performance according to the basic construction method, pull-out resistance, bearing capacity, and settlement amount were measured. At the same diameter, pull-out resistance increased with increasing penetration depth. Simplified comparison is difficult due to the difference in reinforcement method, diameter, and penetration depth, but it showed high bearing capacity in the order of crushed stone pile, PF method, and wood pile foundation. In the case of wood piles, the increase in uplift resistance was different depending on the slenderness ratio. Wood, crushed stone pile and PF construction methods, which are foundation reinforcement works with a bearing capacity of 105 kN/㎡ to 826 kN/㎡, are considered sufficient methods to be applied to the greenhouse foundation. There was a limitation in grasping the consistent trend of each foundation reinforcement method through existing studies. If these data are supplemented through additional empirical tests, it is judged that a basic design guideline that can satisfy the structure and economic efficiency of the greenhouse can be presented.

• Journal of Intelligence and Information Systems
• /
• v.29 no.1
• /
• pp.249-263
• /
• 2023

Collaborative Filtering, a representative recommendation system methodology, consists of two approaches: neighbor methods and latent factor models. Among these, the latent factor model using matrix factorization decomposes the user-item interaction matrix into two lower-dimensional rectangular matrices, predicting the item's rating through the product of these matrices. Due to the factor vectors inferred from rating patterns capturing user and item characteristics, this method is superior in scalability, accuracy, and flexibility compared to neighbor-based methods. However, it has a fundamental drawback: the need to reflect the diversity of preferences of different individuals for items with no ratings. This limitation leads to repetitive and inaccurate recommendations. The Adaptive Deep Latent Factor Model (ADLFM) was developed to address this issue. This model adaptively learns the preferences for each item by using the item description, which provides a detailed summary and explanation of the item. ADLFM takes in item description as input, calculates latent vectors of the user and item, and presents a method that can reflect personal diversity using an attention score. However, due to the requirement of a dataset that includes item descriptions, the domain that can apply ADLFM is limited, resulting in generalization limitations. This study proposes a Generalized Adaptive Deep Latent Factor Recommendation Model, G-ADLFRM, to improve the limitations of ADLFM. Firstly, we use item ID, commonly used in recommendation systems, as input instead of the item description. Additionally, we apply improved deep learning model structures such as Self-Attention, Multi-head Attention, and Multi-Conv1D. We conducted experiments on various datasets with input and model structure changes. The results showed that when only the input was changed, MAE increased slightly compared to ADLFM due to accompanying information loss, resulting in decreased recommendation performance. However, the average learning speed per epoch significantly improved as the amount of information to be processed decreased. When both the input and the model structure were changed, the best-performing Multi-Conv1d structure showed similar performance to ADLFM, sufficiently counteracting the information loss caused by the input change. We conclude that G-ADLFRM is a new, lightweight, and generalizable model that maintains the performance of the existing ADLFM while enabling fast learning and inference.