• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.485-487
• /
• 2023

컴퓨터 비전의 하위 태스크(Task)인 의미론적 분할(Semantic Segmentation)은 자율주행, 해상에서 선박찾기 등 다양한 분야에서 연구되고 있다. 기존 FCN(Fully Conovlutional Networks) 기반 의미론적 분할 모델은 다운샘플링(Dowsnsampling)과정에서 공간정보의 손실이 발생하여 정확도가 하락했다. 본 논문에서는 공간정보 손실을 완화하고자 PSA(Polarized Self-attention)의 공간정보 강조 모듈을 HRNet(High-resolution Networks)의 합성곱 블록 사이에 추가한다. 실험결과 파라미터는 3.1M, GFLOPs는 3.2G 증가했으나 mIoU는 0.26% 증가했다. 공간정보가 의미론적 분할 정확도에 영향이 미치는 것을 확인했다.

• Journal of the Korean Society for Library and Information Science
• /
• v.40 no.3
• /
• pp.175-190
• /
• 2006

For the recent several years, there has been much interest in lexical semantic network However it seems to be very difficult to evaluate the effectiveness and correctness of it and invent the methods for applying it into various problem domains. In order to offer the fundamental ideas about how to evaluate and utilize lexical semantic networks, we developed two automatic vol·d clustering systems, which are called system A and system B respectively. 68.455.856 words were used to learn both systems. We compared the clustering results of system A to those of system B which is extended by the lexical-semantic network. The system B is extended by reconstructing the feature vectors which are used the elements of the lexical-semantic network of 3.656 '-ha' verbs. The target data is the 'multilingual Word Net-CoroNet'. When we compared the accuracy of the system A and system B, we found that system B showed the accuracy of 46.6% which is better than that of system A. 45.3%.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.3
• /
• pp.289-300
• /
• 2010

Recently, diverse information which are location, call history, SMS history, photographs, and video can be collected constantly from mobile devices such as cellular phone, smart phone, and PDA. There are many researchers who study services for searching and abstraction of personal daily life with contextual information in mobile environment. In this paper, we introduce MyLifeBrowser which is developed in our previous work. Also, we explain LPS and correction of GPS coordinates as extensions of previous work and show LPS performance test and evaluate the performance of expanded keywords. MyLifeBrowser which provides searching personal information in mobile device and support of detecting related information according to a fragmentary keyword and common knowledge in ConceptNet. It supports the functionality of searching related locations using Bayesian network that is designed by the authors. In our experiment, we visualize real data through MyLifeBrowser and show the feasibility of LPS server and expanded keywords using both Bayesian network and ConceptNet.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.21 no.2
• /
• pp.198-205
• /
• 2011

GPS, ARPA, AIS, NAVTEX, VHF as modern aids-to-navigation equipments improve the safe navigation and help to reach a reduction in marine accidents by providing images, numeric values, texts, audio-based information for mates, However, we also noticed that it's complicate and difficult for a mate to acquire and analyze such information from these devices while he should devote himself to bridge watchkeeping especially in the urgent situation. Language is another way to get information and free the eyes and hands, so, to solve the problem above, we are trying to propose a new aids-to-navigation system, which can understand and merge multimedia marine safety information, analyze the situation and provide the necessary information in language. In this paper, we try to suggest a meaning interpretation model for supporting linguistic navigation safety information.

• Journal of the Korean Society of Earth Science Education
• /
• v.9 no.1
• /
• pp.65-87
• /
• 2016

The purpose of this study is to examine Korean journals related to science education in order to analyze research trends into Nature of science in Korea. The subject of the study is the level of Korean Citation Index (KCI-listed, KCI listing candidates), that can be searched by the key phrase, "Nature of science" in Korean language through the RISS service. In this study, the Descriptive Statistical Analysis Method is utilized to discover the number of research articles, classifying them by year and by journal. Also, the Sementic Network Analysis was conducted to Word Cloud Analysis the frequency of key words, Centrality Analysis, co-occurrence and Cluster Dendrogram Analysis throughout a variety of research articles. The results show that 91 research papers were published in 25 journals from 1991 to 2015. Specifically, the 2 major journals published more than 50% of the total papers. In relation to research fields., In addition, key phrases, such as 'Analysis', 'recognition', 'lessons', 'science textbook', 'History of Science' and 'influence' are the most frequently used among the research studies. Finally, there are small language networks that appear concurrently as below: [Nature of science - high school student - recognize], [Explicit - lesson - effect], [elementary school - science textbook - analysis]. Research topic have been gradually diversified. However, many studies still put their focus on analysis and research aspects, and there have been little research on the Teaching and learning methods.

• Korean journal of medical education
• /
• v.30 no.4
• /
• pp.317-326
• /
• 2018

Purpose: The study examines changes in students' self-assessment of their general communication (GC) and medical communication (MC) competencies, as well as perceptions of MC concepts. Methods: Participants included 108 second year medical students enrolled at a Korean medical school studying an MC curriculum. It was divided into three sections, and participants responded to questionnaires before and after completing each section. To assess perceived GC and MC competency, items based on a 7-point Likert scale were employed; a single open-ended item was used to examine students' perceptions of MC. Statistical analysis was conducted to gauge GC and MC competency, whereas semantic network analysis was used to investigate students' perceptions of MC. Results: Students perceived their GC competency to be higher than MC. Perceived MC competency differed significantly across the three sections, whereas no differences were found for GC. There were no statistically significant differences after completing the curriculum's second and third sections; however, the vocabulary students used to describe MC concepts became more scholarly and professional. In the semantic networks, the link structure between MC-related words decreased in linearity and looseness, becoming more complex and clustered. The words 'information' and 'transfer' proved integral to students' perceptions; likewise, 'empathy' and 'communication' became closely connected in a single community from two independent communities. Conclusion: This study differed from prior research by conducting an in-depth analysis of changes in students' perceptions of MC, and its findings can be used to guide curriculum development.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.5
• /
• pp.1778-1797
• /
• 2021

Document image binarization is an important pre-processing step in document analysis and archiving. The state-of-the-art models for document image binarization are variants of encoder-decoder architectures, such as FCN (fully convolutional network) and U-Net. Despite their success, they still suffer from three limitations: (1) reduced feature map resolution due to consecutive strided pooling or convolutions, (2) multiple scales of target objects, and (3) reduced localization accuracy due to the built-in invariance of deep convolutional neural networks (DCNNs). To overcome these three challenges, we propose an improved semantic segmentation model, referred to as DP-LinkNet, which adopts the D-LinkNet architecture as its backbone, with the proposed hybrid dilated convolution (HDC) and spatial pyramid pooling (SPP) modules between the encoder and the decoder. Extensive experiments are conducted on recent document image binarization competition (DIBCO) and handwritten document image binarization competition (H-DIBCO) benchmark datasets. Results show that our proposed DP-LinkNet outperforms other state-of-the-art techniques by a large margin. Our implementation and the pre-trained models are available at https://github.com/beargolden/DP-LinkNet.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.1
• /
• pp.245-265
• /
• 2022

In response to problems such as insufficient extraction information, low detection accuracy, and frequent misdetection in the field of Thangka image defects, this paper proposes a YOLOv5 prediction algorithm fused with the attention mechanism. Firstly, the Backbone network is used for feature extraction, and the attention mechanism is fused to represent different features, so that the network can fully extract the texture and semantic features of the defect area. The extracted features are then weighted and fused, so as to reduce the loss of information. Next, the weighted fused features are transferred to the Neck network, the semantic features and texture features of different layers are fused by FPN, and the defect target is located more accurately by PAN. In the detection network, the CIOU loss function is used to replace the GIOU loss function to locate the image defect area quickly and accurately, generate the bounding box, and predict the defect category. The results show that compared with the original network, YOLOv5-SE and YOLOv5-CBAM achieve an improvement of 8.95% and 12.87% in detection accuracy respectively. The improved networks can identify the location and category of defects more accurately, and greatly improve the accuracy of defect detection of Thangka images.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.12
• /
• pp.3364-3382
• /
• 2023

Remote sensing image segmentation plays an important role in realizing intelligent city construction. The current mainstream segmentation networks effectively improve the segmentation effect of remote sensing images by deeply mining the rich texture and semantic features of images. But there are still some problems such as rough results of small target region segmentation and poor edge contour segmentation. To overcome these three challenges, we propose an improved semantic segmentation model, referred to as MRU-Net, which adopts the U-Net architecture as its backbone. Firstly, the convolutional layer is replaced by BasicBlock structure in U-Net network to extract features, then the activation function is replaced to reduce the computational load of model in the network. Secondly, a hybrid multi-scale recognition module is added in the encoder to improve the accuracy of image segmentation of small targets and edge parts. Finally, test on Massachusetts Buildings Dataset and WHU Dataset the experimental results show that compared with the original network the ACC, mIoU and F1 value are improved, and the imposed network shows good robustness and portability in different datasets.

• Journal of Digital Convergence
• /
• v.19 no.1
• /
• pp.307-314
• /
• 2021

This study aimed to apply the principle of the semantic network to a long novel in an attempt to understand the structure of the entire document and the manifested relationships between words and words. The costume expressions in Murakami's novel Norwegian Wood were analyzed based on the characters' symbols, relationships, and personality characteristics. The study identified the symbols of the characters in the novel and the relationship properties between the characters through the Clauset-Newman-Moore clustering algorithm. The descriptions and symbols of the relationships between the characters were identified within the worldview that the author had intended. Further, it was confirmed that the expression of each costume according to the character's personality was also connected to the clue that explained said character. This fusion study is academically significant in that it presents a new methodology for analyzing literary works