• 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 Korea Multimedia Society
• /
• v.25 no.3
• /
• pp.493-501
• /
• 2022

This paper is a study on context-sensitive spelling error correction and uses the Korean WordNet (KorLex)[1] that defines the relationship between words as a graph to improve the performance of the correction[2] based on the vector information of the word embedded in the correction technique. The Korean WordNet replaced WordNet[3] developed at Princeton University in the United States and was additionally constructed for Korean. In order to learn a semantic network in graph form or to use it for learned vector information, it is necessary to transform it into a vector form by embedding learning. For transformation, we list the nodes (limited number) in a line format like a sentence in a graph in the form of a network before the training input. One of the learning techniques that use this strategy is Deepwalk[4]. DeepWalk is used to learn graphs between words in the Korean WordNet. The graph embedding information is used in concatenation with the word vector information of the learned language model for correction, and the final correction word is determined by the cosine distance value between the vectors. In this paper, In order to test whether the information of graph embedding affects the improvement of the performance of context- sensitive spelling error correction, a confused word pair was constructed and tested from the perspective of Word Sense Disambiguation(WSD). In the experimental results, the average correction performance of all confused word pairs was improved by 2.24% compared to the baseline correction performance.

• The Journal of the Korea Contents Association
• /
• v.21 no.11
• /
• pp.478-489
• /
• 2021

The mobile game industry is currently on the rise and is one of the major industries. Recently, advertisements among mobile games are aimed at branded content, unlike existing advertising content. In this study, we conduct a study on the advertisement effectiveness of the attributes of branded advertisements under a mobile game environment. Therefore, the impact of context congruity, perceived advertising intrusiveness and entertainment on advertising attitude, brand attitude and purchasing intention was verified through experimental research on Chinese mobile game users in the 10s and 20s. Analysis of the PLS structural equation model showed that context congruity and entertainment had a significant impact on advertising attitude, but did not have a significant impact on brand attitude. Perceived advertising intrusiveness has no significant impact on advertising attitude and brand attitude. Finally, advertising attitude had a significant influence on purchase intention through brand attitude. Through these findings, we discuss the effectiveness of mobile game branded advertisement and present effective brand advertising strategies.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.8
• /
• pp.1231-1237
• /
• 2022

It is necessary to dynamic recognition system with real time loading height and pressure of the loading waste, the drying of wood, batteries, and plastic wastes, which are representative compositional wastes, and the carbonization changes on the surface. The dynamic context awareness service constituted a platform based on Universal Middleware system using BCN convergence communication service as a Ambient SDK model. A context awareness system should be constructed to determine the cause of the fire based on the analysis data of fermentation heat point with natural ignition from the load waste. Furthermore, a real-time dynamic service platform that could be apply to the configuration of scenarios for each type from early warning fire should be built using Universal Middleware. Thus, this issue for Internet of Things realize recognition platform for analyzing low temperature fired fire possibility data should be dynamically configured and presented.

• Korean Journal of Remote Sensing
• /
• v.38 no.5_2
• /
• pp.685-694
• /
• 2022

Building information is essential for various urban spatial analyses. For this reason, continuous building monitoring is required, but it is a subject with many practical difficulties. To this end, research is being conducted to extract buildings from satellite images that can be continuously observed over a wide area. Recently, deep learning-based semantic segmentation techniques have been used. In this study, a part of the structure of the context-based ResU-Net was modified, and training was conducted to automatically extract a building from a 30 cm Worldview-3 RGB image using SpaceNet's building v2 free open data. As a result of the classification accuracy evaluation, the f1-score, which was higher than the classification accuracy of the 2nd SpaceNet competition winners. Therefore, if Worldview-3 satellite imagery can be continuously provided, it will be possible to use the building extraction results of this study to generate an automatic model of building around the world.

• International Journal of Computer Science & Network Security
• /
• v.22 no.11
• /
• pp.11-18
• /
• 2022

Linking Ukraine's education system with the trends of global digitalization is mandatory to ensure the sustainable, long-term development of the country, as well as to increase the sustainability of the education system and the economy as a whole during the crisis period. Now the main problems of the education system in Ukraine are manifested in a complex context caused by Russian armed aggression. In the context of war, problems include differences in adaptation to online learning among educational institutions, limited access to education for vulnerable groups in the zone of active hostilities, the lack of digital educational resources suitable for online learning, and the lack of basic digital skills and competencies among students and teachers necessary to properly conduct online classes. Some of the problems of online learning were solved in the pandemic, but in the context of war Ukrainian society needs a new vision of education and continuous efforts of all social structures in the public and private environment. In the context of war, concerted action is needed to keep education on track and restore it in active zones, adapting to the needs of a dynamic society and an increasingly digitized economy. Among the urgent needs of the education system are a change in the teaching-learning paradigm, which is based on content presentation, memorization, and reproduction, and the adoption of a new, hybrid educational model that will encourage the development of necessary skills and abilities for students and learners in a digitized society and enable citizens close to war zones to learn.

• Culinary science and hospitality research
• /
• v.22 no.1
• /
• pp.87-95
• /
• 2016

The purpose of current study is to develop and estimate a proposed model that explains the potential relationships among food quality, experiential value, price fairness, image, satisfaction, and behavioral intention in context of water park. In addition, the study will verify how these factors link to each other. Results show that food quality is a significant antecedent of experience value, price fairness, water park image. Also, the experiential value and water park image influence on visitors' satisfaction. Last, the satisfaction is critical predictor of behavioral intention. These findings will contribute to understand the consumers' perception about water park, and how derives the customer satisfaction and behavioral intention. In sum, present study will serve insights for industry marketers and managers in water park segment.

• Culinary science and hospitality research
• /
• v.20 no.5
• /
• pp.93-100
• /
• 2014

This study performed to identify the relationships among experiential value, brand image, satisfaction and customer loyalty in context of full-service restaurant, and to find the moderating effect of gender on the formulated model. SPSS 18.0 and AMOS 18.0 were employed to conduct frequency analysis, reliability analysis, exploratory and confirmatory factor analysis, and multigroup analysis to examine moderating effect. Results confirmed the validity and reliability and found significant relationships among the constructs. First, two factors of experiential value (e.g., aesthetic and economic value) have positive influence on brand image, satisfaction, and brand image was significant predictor of customer satisfaction. Second, satisfaction was significant antecedent of attitudinal loyalty and the attitudinal loyalty has influence on behavioral loyalty. In addition, current study identified moderating effect of gender between playfulness and brand image even though there was on significant relationship between both constructs. These results will be meaningful for developing marketing strategies and successful business especially for full-service restaurants.

• MALSORI
• /
• v.55
• /
• pp.103-118
• /
• 2005

In this paper, we describe a morpheme-based pronunciation lexicon useful for Korean LVCSR. The phonemic-context-dependent multiple pronunciation lexicon improves the recognition accuracy when cross-morpheme pronunciation variations are distinguished from within-morpheme pronunciation variations. Since adding all possible pronunciation variants to the lexicon increases the lexicon size and confusability between lexical entries, we have developed a lexicon pruning scheme for optimal selection of pronunciation variants to improve the performance of Korean LVCSR. By building a proposed pronunciation lexicon, an absolute reduction of $0.56\%$ in WER from the baseline performance of $27.39\%$ WER is achieved by cross-morpheme pronunciation variations model with a phonemic-context-dependent multiple pronunciation lexicon. On the best performance, an additional reduction of the lexicon size by $5.36\%$ is achieved from the same lexical entries.

• ETRI Journal
• /
• v.40 no.4
• /
• pp.483-498
• /
• 2018

Nowadays, with progresses in robotic science, the design and implementation of a mechanism for human-robot interaction with a low workload is inevitable. One notable challenge in this field is the interaction between a single human and a group of robots. Therefore, we propose a new comprehensive framework for single-human multiple-robot remote interaction that can form an efficient intelligent adaptive interaction (IAI). Our interaction system can thoroughly adapt itself to changes in interaction context and user states. Some advantages of our devised IAI framework are lower workload, higher level of situation awareness, and efficient interaction. In this paper, we introduce a new IAI architecture as our comprehensive mechanism. In order to practically examine the architecture, we implemented our proposed IAI to control a group of unmanned aerial vehicles (UAVs) under different scenarios. The results show that our devised IAI framework can effectively reduce human workload and the level of situation awareness, and concurrently foster the mission completion percentage of the UAVs.