• Korean Journal of Remote Sensing
• /
• v.39 no.2
• /
• pp.143-156
• /
• 2023

To precisely and reliably analyze the contents of the satellite imagery, recognizing the clouds which are the obstacle to gathering the useful information is essential. In recent times, deep learning yielded satisfactory results in various tasks, so many studies using deep neural networks have been conducted to improve the performance of cloud detection. However, existing methods for cloud detection have the limitation on increasing the performance due to the adopting the network models for semantic image segmentation without modification. To tackle this problem, we introduced the multi-branch neural architecture search to find optimal network structure for cloud detection. Additionally, the proposed method adopts the soft intersection over union (IoU) as loss function to mitigate the disagreement between the loss function and the evaluation metric and uses the various data augmentation methods. The experiments are conducted using the cloud detection dataset acquired by Arirang-3/3A satellite imagery. The experimental results showed that the proposed network which are searched network architecture using cloud dataset is 4% higher than the existing network model which are searched network structure using urban street scenes with regard to the IoU. Also, the experimental results showed that the soft IoU exhibits the best performance on cloud detection among the various loss functions. When comparing the proposed method with the state-of-the-art (SOTA) models in the field of semantic segmentation, the proposed method showed better performance than the SOTA models with regard to the mean IoU and overall accuracy.

• Journal of Digital Convergence
• /
• v.16 no.8
• /
• pp.161-167
• /
• 2018

The purpose of this study is to investigate how the advertising campaign is spreading in social networks, and how the advertising model plays an important role in advertisement diffusion. In order to grasp the diffusion patterns of advertising, a text mining and social network analysis were conducted using the beer brand 'Kloud' as a collection keyword. After analyzing the social data for two months since the on-air of 'Good Body' advertisement, which was the first ad that "Sulhyun" appeared in. After the launch of the ad, Kloud has been mainly associated with keywords such as 'yavis & trendy style', 'beer brand', 'beer matching food', 'luxury beer drinking place', 'leisure trend', and 'SNS activity', etc. In addition, "Sul Hyun" also showed that an advertising model contributes to the spread of advertisement on social media in terms of image transition as well as brand's name and unique selling point.

• The Journal of the Korea Contents Association
• /
• v.20 no.2
• /
• pp.240-247
• /
• 2020

There have been many efforts to understand the trends of IT environments that have been rapidly changed. In a view point of management, it needs to prepare the social systems in advance by using Big-data these days. This research is for the implementation of Issue Analysis System for the Big-data based on Artificial Intelligence. This paper aims to confirm the possibility of new technology for Big-data processing through the proposed Issue Analysis System using. We propose a technique for semantic reasoning and pattern analysis based on the AI and show the proposed method is feasible to handle the Big-data. We want to verify that the proposed method can be useful in dealing with Big-data by applying latest security issues into the system. The experiments show the potentials for the proposed method to use it as a base technology for dealing with Big-data for various purposes.

• Journal of KIISE:Databases
• /
• v.29 no.6
• /
• pp.492-501
• /
• 2002

An asynchronous trigger processor (ATP) is a oftware system that processes triggers after update transactions to databases are complete. In an ATP, discrimination networks are used to check the trigger conditions efficiently. Discrimination networks store their internal states in memory nodes. TriggerMan is an ATP and uses Gator network as the .discrimination network. The changes in databases are delivered to TriggerMan in the form of tokens. Processing tokens against a Gator network updates the memory nodes of the network and checks the condition of a trigger for which the network is built. Parallel token processing is one of the methods that can improve the system performance. However, uncontrolled parallel processing breaks trigger processing semantic consistency. In this paper, we propose four trigger processing consistency levels that allow parallel token processing with minimal anomalies. For each consistency level, a parallel token processing technique is developed. The techniques are proven to be valid and are also applicable to materialized view maintenance.

• Journal of Broadcast Engineering
• /
• v.26 no.4
• /
• pp.419-428
• /
• 2021

Shadow removal plays a key role for the pre-processing of image processing techniques such as object tracking and detection. With the advances of image recognition based on deep convolution neural networks, researches for shadow removal have been actively conducted. In this paper, we propose a novel method for shadow removal, which utilizes self attention distillation to extract semantic features. The proposed method gradually refines results of shadow detection, which are extracted from each layer of the proposed network, via top-down distillation. Specifically, the training procedure can be efficiently performed by learning the contextual information for shadow removal without shadow masks. Experimental results on various datasets show the effectiveness of the proposed method for shadow removal under real world environments.

• Journal of the Architectural Institute of Korea Planning & Design
• /
• v.35 no.12
• /
• pp.83-92
• /
• 2019

The purpose of this study is to analyze the continuous forms of time and space that work as architectural design principles of the Memorial to the Murdered Jews of Europe (Jewish Memorial). Continuity is divided into two, physical and non-physical attributes. The former extends from small installations to furnitures, finishes, spatial composition, and even spatial networks that complete architecture, and the latter is tied to time, from traditional to historical, developmental, commemorative or memorial elements. They are inherent in architecture as continuous forms. The Jewish Memorial is analyzed by the analysing framework of these two items. The analysis of the two layers can be summarized as follows; physical continuity is found in the space networks of the Memorial's inside and outside, the undulating spaces, the finishes, the small furnitures and installations, the entrances of staircases, the expanded underground of the ground order, and non-physical continuity manifests in the topological variation of spatiality, the morphological development of memorial architecture, the connection of semantic sense of space, and the superposition of historicity and modernity. These forms of continuity do not aestheticize the German enormity history, but make the meaning of the Memorial into non-superficial, in-depth architecture as a monument. Thus, the results of this study show that physical and non-physical continuity should be considered as the important design principles of architecture that makes the Memorial architecture possible.

• The Journal of Korean Institute of Information Technology
• /
• v.17 no.7
• /
• pp.31-45
• /
• 2019

The core technologies for intelligent services today are deep learning, that is neural networks, and parallel and distributed processing technologies such as GPU parallel computing and big data. However, for intelligent services and knowledge sharing services through globally shared ontologies in the future, there is a technology that is better than the neural networks for representing and reasoning knowledge. It is a knowledge representation of IF-THEN in RIF or SWRL, which is the standard rule language of the Semantic Web, and can be inferred efficiently using the rete algorithm. However, when the number of rules processed by the rete algorithm running on a single computer is 100,000, its performance becomes very poor with several tens of minutes, and there is an obvious limitation. Therefore, in this paper, we analyze the past and current studies on parallel and distributed processing of rete algorithm, and examine what aspects should be considered to implement an efficient rete algorithm.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.7
• /
• pp.1706-1725
• /
• 2024

The unique U-shaped structure of U-Net network makes it achieve good performance in image segmentation. This network is a lightweight network with a small number of parameters for small image segmentation datasets. However, when the medical image to be segmented contains a lot of detailed information, the segmentation results cannot fully meet the actual requirements. In order to achieve higher accuracy of medical image segmentation, a novel improved U-Net network architecture called multi-scale encoder-decoder U-Net+ (MEDU-Net+) is proposed in this paper. We design the GoogLeNet for achieving more information at the encoder of the proposed MEDU-Net+, and present the multi-scale feature extraction for fusing semantic information of different scales in the encoder and decoder. Meanwhile, we also introduce the layer-by-layer skip connection to connect the information of each layer, so that there is no need to encode the last layer and return the information. The proposed MEDU-Net+ divides the unknown depth network into each part of deconvolution layer to replace the direct connection of the encoder and decoder in U-Net. In addition, a new combined loss function is proposed to extract more edge information by combining the advantages of the generalized dice and the focal loss functions. Finally, we validate our proposed MEDU-Net+ MEDU-Net+ and other classic medical image segmentation networks on three medical image datasets. The experimental results show that our proposed MEDU-Net+ has prominent superior performance compared with other medical image segmentation networks.

• Journal of Internet Computing and Services
• /
• v.13 no.2
• /
• pp.99-108
• /
• 2012

Many methods had been developed to improve the accuracy of extracting information from a vast amount of data. This paper combined a number of natural language processing methods such as NER (named entity recognition), sentence extraction, and part of speech tagging to carry out text analysis. The data source is comprised of texts obtained from the web using a domain-specific data extraction agent. A framework for the extraction of information from unstructured data was developed using the aforementioned natural language processing methods. We simulated the performance of our work in the extraction and analysis of texts for the detection of organizational structures. Simulation shows that our study outperformed other NER classifiers such as MUC and CoNLL on information extraction.

• 한국HCI학회:학술대회논문집
• /
• 2006.02a
• /
• pp.735-740
• /
• 2006

유비쿼터스 가정환경에서 서비스를 제공하기 위한 다양한 장치들은 각기 고유한 인터페이스를 가진다. 사용자는 이 장치들을 제어하기 위해서 각각 다른 인터페이스에 익숙해야 하며, 결국 장치 수만큼의 인터페이스를 다루어야 한다. 이와 같은 불편을 해소하기 위해서는 하나의 입력 장치로 여러 장치들을 조작하는 사용자 인터페이스가 필요하다. 특히 유비쿼터스 가정환경에서는 다양한 장치들의 상태 및 기능 등이 동적으로 변하고, 장치가 설정되는 환경도 일정하지 않기 때문에 사용자 중심의 유비쿼터스 환경을 제공하기 위해서는 다양한 인터페이스를 통합할 필요가 있다. 사용자가 비슷하게 인지하는 이종 장치들의 기능을 통합하여 사용자 인터페이스의 동일한 입력으로 매핑한다면 사용자의 부담을 줄일 수 있을 것이다. 본 논문에서는 유비쿼터스 가정환경의 다양한 장비들과 인터페이스 사이의 입출력 관계를 분석하여 시맨틱 네트워크로 모델링하는 방법을 제안한다. 각 장치의 상태와 기능을 시맨틱 네트워크로 정의하고, 노드나 엣지 사이의 유사도를 평가하여 장치와 사용자 인터페이스 사이를 자동으로 매핑한다. 제안하는 방법을 가정환경 입출력장치에 적용하고, 입출력 매핑을 시뮬레이션하는 환경을 구현하여 유용성을 검증한다.