• Journal of Broadcast Engineering
• /
• v.27 no.6
• /
• pp.897-905
• /
• 2022

Research for video key frame detection has been actively conducted in the fields of computer vision. Recently with the advances on deep learning techniques, performance of key frame detection has been improved, but the various type of video content and complicated background are still a problem for efficient learning. In this paper, we propose a novel method for key frame detection, witch utilizes contrastive learning and memory bank module. The proposed method trains the feature extracting network based on the difference between neighboring frames and frames from separate videos. Founded on the contrastive learning, the method saves and updates key frames in the memory bank, witch efficiently reduce redundancy from the video. Experimental results on video dataset show the effectiveness of the proposed method for key frame detection.

• The Journal of the Korea Contents Association
• /
• v.22 no.3
• /
• pp.261-272
• /
• 2022

The aim of the study is to develop the Adolescent Anger Provocation Scale(AAPS) for Korean adolescents. In order to develop the scale, two research processes were conducted. First, a total of 130 items were derived from various related research. After analyzing 130 items by a research team(2 related field professors and 2 Ph.D. students), a total of 54 items reconstructed based on redundancy, cultural relevancy, and appropriateness. Second, a survey was administered among 252 Korean adolescents. After analyzing an exploratory factor analysis, 12 items of poor factor loading were deleted. Thus, seven factors and 42 items were extracted. Seven factors were as follows: (1) threat to self-esteem, (2) uncontrollable situations, (3) violation of personal property, (4) incidents associated with rejection, (5) violation of regulation or rules, (6) disrespectful treatment, (7) unfair treatment due to age. The internal consistency and convergence validity of the scale were supported. The results of confirmatory factor analysis were also reported. In addition, study limitations and recommendations of future research were discussed.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.12
• /
• pp.77-84
• /
• 2022

In this paper, we propose a Label Reassignment Strategy to improve the performance of an object detection algorithm. Our approach involves two stages: an inference stage and an assignment stage. In the inference stage, we perform multi-scale inference with predefined scale sizes on a trained model and re-infer masked images to obtain robust classification results. In the assignment stage, we calculate the IoU between bounding boxes to remove duplicates. We also check box and class occurrence between the detection result and annotation label to re-assign the dominant class type. We trained the YOLOX-L model with the re-annotated dataset to validate our strategy. The model achieved a 3.9% improvement in mAP and 3x better performance on AP_S compared to the model trained with the original dataset. Our results demonstrate that the proposed Label Reassignment Strategy can effectively improve the performance of an object detection model.

• Journal of Broadcast Engineering
• /
• v.28 no.2
• /
• pp.194-212
• /
• 2023

The MPEG immersive video (MIV) coding standard achieved high compression efficiency by removing inter-view redundancy and merging the residuals of immersive video which consists of multiple texture (color) and geometry (depth) pairs. Grouping of views that represent similar spaces enables quality improvement and implementation of selective streaming, but this has not been actively discussed recently. This paper introduces an implementation of group-based encoding into the recent version of MIV reference software, provides experimental results on optimal views and videos per group, and proposes a decision method for optimal number of videos for global immersive video representation by using portion of residual videos.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.6 no.5
• /
• pp.337-345
• /
• 2016

There is an increasing attention on the sustainable energy in the world as the problem of fossil energy exhaustion and environmental issues appear. This paper addresses technologies and industries related to hydrogen energy in the literature, and develops strategy for the preparation of hydrogen energy infrastructure in Korea by extracting several factors suitable for SWOT analysis after reviewing literature. Research articles and reports about the practical application of hydrogen energy in the literature were investigated and seventy evaluation factors were extracted. Then, survey to eliminate the unsuitability problems of duplication, disproportion and objectivity was performed by scholars and experts in research institutes. Finally, sixteen factors consisting of 4 strength factors, 4 weakness factors, 5 opportunity factors and 4 threat factors were determined as the strategic factors for the infrastructure of hydrogen energy industry.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2020.06a
• /
• pp.386-386
• /
• 2020

가뭄은 강수, 증발산, 대기온도, 토양수분 등 다양한 수문기상학적 인자들이 복합적으로 작용하여 발생되기 때문에 가뭄의 정확한 사상을 분석하는 것은 매우 어렵다. 또한 어떤 요인을 중심으로 고려하느냐에 따라 가뭄은 다양한 시각으로 정의되고 있다. 일정기간 평균 강수량보다 적은 강수로 인해 건조한 날이 지속되는 것, 즉 기상요소를 중심으로 가뭄을 정의하는 것을 기상학적 가뭄이라 하며, 작물의 생육에 필요한 수분을 중심으로 고려하는 것을 농업적 가뭄이라 한다. 또한 하천유량, 댐 저수량 등 전반적인 수자원 공급원의 부족을 수문학적 가뭄이라 한다. 이와 같이 다양하게 나타는 가뭄의 발생특성을 정량적으로 해석하기 위해 다양한 가뭄지수가 개발되어 왔다. 그러나 현재까지 개발된 가뭄지수들은 공통적으로 정형데이터를 활용하여 산정한다. 하지만 최근에는 비정형데이터를 활용하여 지수(Index)를 산정하거나, 재난관리에 적용하는 등 비정형 데이터의 활용이 급증하고 있다. 따라서 본 연구에서는 비정형 데이터(뉴스 데이터)를 활용하여 가뭄지수를 산정하고 기존의 가뭄지수들과의 상관성 분석을 실시 한 뒤, 지수결합을 통해 가뭄사상 분석의 새로운 방안을 제시하고자 하였다. 본 연구의 공간적범위는 2014~2015 충남서북부가뭄 지역 중 가장 큰 피해를 입었던 보령지역으로 선정하였으며 시간적범위는 2013~2016년으로 설정하였다. 비정형 데이터의 구축은 크롤링(Crawling)을 활용하여 네이버 뉴스의 기사를 수집하였으며 자료의 신뢰성을 위해 URL이 동일한 중복기사 및 '보령', '가뭄' 단어가 없는 기사는 제거하였다. 구축된 데이터를 기반으로 월별 빈도를 산출하고 표준점수(Z-score)로 환산하여 가뭄지수를 산정하였다. 산정된 가뭄지수가 어떤 가뭄의 유형(기상학적, 농업적, 수문학적)을 보이는지 확인하기 위해 기존의 가뭄지수들과 상관성분석을 실시하였으며, 가장 높은 상관성을 보이는 가뭄지수와 결합을 통해 새로운 가뭄 사상을 분석하였다. 본 연구에서 진행한 가뭄사상 분석은 향후 가뭄만이 아니라 다양한 재난분야에서 비정형 데이터를 활용한 분석의 기초로자료로 활용될 수 있을 것이다.

• Journal of the Korean Society for Library and Information Science
• /
• v.57 no.4
• /
• pp.97-117
• /
• 2023

The purpose of this study is to develop descriptive metadata elements of contents of oral history records. First, the meaning of contents in transcripts and detailed interview lists, as well as the characteristics of metadata for oral history records, were examined. Second, elements related contents of domestic and international oral history metadata were compared to identify 18 elements, then streamlined 12 elements were set by removing duplicates. Third, 12 elements were compared with detailed interview lists based on transcripts of 6 interviewees who had interviews on knit business in Haebangchon. The result of this study discovered 1 additional element that designated as 'basic' and these were further categorized into 15 element of 'enhanced'. These elements are expected to contribute to expanding the scope of utilization of oral history records by supporting the structure and expression of that contents.

• Smart Media Journal
• /
• v.12 no.11
• /
• pp.77-85
• /
• 2023

After COVID-19, communication through online platforms has increased, leading to an accumulation of massive amounts of conversational text data. With the growing importance of summarizing this text data to extract meaningful information, there has been active research on deep learning-based abstractive summarization. However, conversational data, compared to structured texts like news articles, often contains missing or transformed information, necessitating consideration from multiple perspectives due to its unique characteristics. In particular, vocabulary omissions and unrelated expressions in the conversation can hinder effective summarization. Therefore, in this study, we restructured by considering the characteristics of Korean conversational data, fine-tuning a pre-trained text summarization model based on KoBART, and improved conversation data summary perfomance through a refining operation to remove redundant elements from the summary. By restructuring the sentences based on the order of utterances and extracting a central speaker, we combined methods to restructure the conversation around them. As a result, there was about a 4 point improvement in the Rouge-1 score. This study has demonstrated the significance of our conversation restructuring approach, which considers the characteristics of dialogue, in enhancing Korean conversation summarization performance.

• Journal of the Korean Society for information Management
• /
• v.41 no.3
• /
• pp.85-107
• /
• 2024

This study analyzes trends in data curation research. A total of 1,849 scholarly records were extracted from Scopus and WoS, with 1,797 papers selected after removing duplicates. Titles, keywords, and abstracts were analyzed through keyword frequency analysis, LDA topic modeling, and network analysis. Frequent keywords like 'research' and 'information' suggest that data curation is widely applied in medical research, biomedical research, data management, and infrastructure. LDA modeling identified five main topics: improving medical data quality, enhancing big data management, managing scientific data and repositories, annotating and modeling medical data, and gene/protein database research. Network analysis showed that 'analysis' was central in global discussions, while 'gene' and 'system' were locally central. These findings highlight the importance of data curation in various research areas.

• Journal of Society of Preventive Korean Medicine
• /
• v.28 no.2
• /
• pp.85-97
• /
• 2024

목적 : 본 연구는 한의약에서 사용하는 용어가 SNOMED CT로 매핑 가능한지 여부를 조사하고, 한의약 용어를 표현하기 위해 기존 SNOMED CT 온톨로지를 개선할 수 있는 방안을 제안하는 것을 목표로 하였다. 방법 : 선행 연구의 매핑 가이드라인에서 제시된 7단계 과정을 수정하여 활용하였다. 매핑의 목적 및 범위 정의, 용어 추출, 개념 추출, 매핑을 위한 소스 용어 작업, SNOMED CT 개념 검색, 매핑 관계 분류 및 매핑 검증의 과정을 수행하였다. 매핑의 목적은 한의약 임상 아이디어를 표현하는 표준 용어로서 SNOMED CT를 평가하는 것이고, 범위에는 편두통 환자 관리의 평가, 진단, 치료 및 예방을 포함하였다. 결과 : 총 546개의 용어가 추출되었다. 중복된 용어를 제거한 후, 271개의 개념이 SNOMED CT 매핑에 사용되었다. 이중 43.2%는 SNOMED CT 개념과 의미론적으로 동등하게 매핑되었고(117개 개념), 39.1%는 SNOMED CT 개념이 더 포괄적인 의미를 가지도록 매핑되었다(106개 개념). 상대적으로 포괄적인 의미를 가지는 SNOMED CT 개념에 매핑된 한의약 개념 106개 중 19개는 SNOMED CT 후조합을 이용하여 의미론적으로 동등하게 표현이 가능하였다. 나머지 17.7%의 한의약 개념은 SNOMED CT에 매핑할 수 없었다. 결론 : 본 연구는 한의약에서 사용되는 개념을 SNOMED CT에 매핑하여 한의약 용어를 표준화하였다. 연구 결과를 바탕으로, 한의약에서 사용되는 용어를 표준의료용어로 표현하기 위하여 SNOMED CT에 새로운 개념과 속성을 추가하는 것을 제안한다.