• Journal of KIISE:Software and Applications
• /
• v.31 no.10
• /
• pp.1386-1399
• /
• 2004

This paper presents correlation-based automatic image captioning. Given a training set of annotated images, we want to discover correlations between visual features and textual features, so that we can automatically generate descriptive textual features for a new unseen image. We develop models with multiple design alternatives such as 1) adaptively clustering visual features, 2) weighting visual features and textual features, and 3) reducing dimensionality for noise sup-Pression. We experiment thoroughly on 10 data sets of various content styles from the Corel image database, about 680MB. The major contributions of this work are: (a) we show that careful weighting visual and textual features, as well as clustering visual features adaptively leads to consistent performance improvements, and (b) our proposed methods achieve a relative improvement of up to 45% on annotation accuracy over the state-of-the-art, EM approach.

• Journal of Digital Contents Society
• /
• v.19 no.8
• /
• pp.1463-1469
• /
• 2018

The development of multimedia devices with built-in cameras such as smart devices and various studies using video-related multimedia data such as images and video obtained from the devices have been actively conducted. These studies deal with image data. An image can be defined as a set of color information obtained from a digital sensor called a pixel. Images contain various cognitive information such as color, lighting, objects and so on. In order to extract or process such information, it is necessary to clearly understand the composition of colors. In this paper, we introduce 3-dimensional information visualization method which can effectively express the results of image processing together with color distribution. This study visualizes the characteristics of image related multimedia data as well as the characteristics of various analytical data derived from it, so that researchers can transmit the image information more clearly and effectively.

• Journal of The Korean Association For Science Education
• /
• v.41 no.5
• /
• pp.371-390
• /
• 2021

The purpose of this study is to develop a neuropsychological model for the spatial ability factor and to divide the brain active area involved in the light & shadow problem solving process into the domain-general ability and the domain-specific ability based on the neuropsychological model. Twenty-four male college students participated in the study to measure the synchronized eye movement and electroencephalograms (EEG) while they performed the spatial ability test and the light & shadow tasks. Neuropsychological model for the spatial ability factor and light & shadow problem solving process was developed by integrating the measurements of the participants' eye movements, brain activity areas, and the interview findings regarding their thoughts and strategies. The results of this study are as follows; first, the spatial visualization and mental rotation factors mainly required activation of the parietal lobe, and the spatial orientation factor required activation of the frontal lobe. Second, in the light & shadow problem solving process, participants use both their spatial ability as a domain-general thought, and the application of scientific principles as a domain-specific thought. The brain activity patterns resulting from a participants' inferring the shadow by parallel light source and inferring the shadow when the direction of the light changed were similar to the neuropsychological model for the spatial visualization factor. The brain activity pattern from inferring an object from its shadow by light from multiple directions was similar to the neuropsychological model for the spatial orientation factor. The brain activity pattern from inferring a shadow with a point source of light was similar to the neuropsychological model for the spatial visualization factor. In addition, when solving the light & shadow tasks, the brain's middle temporal gyrus, precentral gyrus, inferior frontal gyrus, middle frontal gyrus were additionally activated, which are responsible for deductive reasoning, working memory, and planning for action.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.9
• /
• pp.841-849
• /
• 2008

Natural disaster such as inundation due to the typhoon induced storm-surge has inflicted severe losses on the coastal area. The problem of global warming and sea surface rising has issued and thus influences the increase of frequency and potential power of storm-surge. What we can do is to make intelligent effort to predict and prevent the losses through the early warning and prevention activity from the accurate prediction and forecasting about the time-varying storm-surge height and its arriving time resulted from the numerical simulation with sea observations. In this paper, we developed the web service based GIS-Enabled storm-surge visualization system to predict and prevent the storm-surge disasters. Moreover. for more accurate topography around coastal area and fine-grid storm-surge numerical model, we have accomplished GIS-based coastal mapping through LiDAR measurement.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.16 no.6
• /
• pp.4170-4175
• /
• 2015

The global warming has been recognized as a serious problem. To solve this problem, the country of all the world have conducted techniques such as energy saving and reduction of greenhouse gas. The 30% of total energy is being consumed by buildings domestically and BAS/BEMS are being operated for the management of building energy. But, BEMS is unsatisfactory condition to manage the energy efficiently. In this study, We investigate real application on Chamsaem elementary school as a target buildings and develop 3D Web Browser platform. The spatial information of target buildings is described with used BIM and the energy information is presented for facilities operation and electricity consumption in floor and facility. We perform the quantitative evaluation based on ISO/IEC 9126 and operate this system for two months. in result, energy decreased by about 16%. We expect that the proposed 3D Web Browser system will bring operating efficiency improvement of the current BEMS.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2007.11a
• /
• pp.440-447
• /
• 2007

사진은 시각화되어 있으면서도 사용자가 생성하기 쉽다는 이유로 가장 인기 있는 컨텐트 중 하나이며, 사진을 찍은 위치 정보는 해당 장소가 갖는 특징 혹은 상징성으로, 상품 및 서비스와 연관성을 갖고 있기 때문에 상거래의 매개체가 될 가능성을 내포하고 있다. 그러나 현재의 상거래 환경에서는 사진의 위치정보를 자동화, 체계화하여 저장하고 활용하는데 한계를 가지고 있어서 사진 자체가 상거래에 활용되는 모델을 찾아보기 힘들다. 본 연구에서 제시하는 U-Photo 비즈니스 모델은 사용자가 찍은 사진의 배경에 해당하는 장소를 그 장소를 통해 광고를 하고자 하는 광고주와 연계하고, 그 사진을 클릭했을 때 광고주의 사이트가 로딩 되도록 하는, 컨텐트 생성자, 컨텐트 소비자, 광고주 연계 비즈니스 모델이다. 본 논문은 유비쿼터스 컴퓨팅 환경에서 사진을 활용한 비즈니스 모델을 제안하고, 본 비즈니스 모델이 어떤 함의를 지니고 있는지를 분석하며 시장에서 실제 작동할 조건을 예측하여 본 비즈니스 모델을 평가한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.67-70
• /
• 2022

NLP 분야에서 적극 활용되기 시작한 Transformer 모델을 Vision 분야에서 적용하기 시작하면서 object detection과 segmentation 등 각종 분야에서 기존 CNN 기반 모델의 정체된 성능을 극복하며 향상되고 있다. 또한, label 데이터 없이 이미지들로만 자기지도학습을 한 ViT(Vision Transformer) 모델을 통해 이미지에 포함된 여러 중요한 객체의 영역을 검출하는 saliency map을 추출할 수 있게 되었으며, 이로 인해 ViT의 자기지도학습을 통한 object detection과 semantic segmentation 연구가 활발히 진행되고 있다. 본 논문에서는 ViT 모델 뒤에 classifier를 붙인 모델에 일반 학습한 모델과 자기지도학습의 pretrained weight을 사용해서 전이학습한 모델의 시각화를 통해 각 saliency map들을 비교 분석하였다. 이를 통해, 클래스 분류 학습 기반 전이학습이 transformer의 saliency map에 미치는 영향을 확인할 수 있었다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.12 no.5
• /
• pp.424-431
• /
• 2002

This paper proposes a new quantitative image assessment model which is essential to verify the performance of block-based DCT coding. The proposed model considers not only global distortions such as frequency sensitivity and channel masking using HVS based visual model, but also distortions including several local distortions caused by block-based coding.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.475-478
• /
• 2023

컴퓨터 단층촬영(CT)을 활용한 골격근 단면적은 근감소증과 관련된 기능을 평가하는 데 사용된다. 일반적인 근감소증 연구는 요추 3번의 골격근량을 주로 보지만 암 또는 폐절제술과의 상관관계를 예측하기 위한 다양한 연구에서는 흉추 4번, 7번, 8번, 10번, 12번 다양한 수준의 골격근량으로 연구를 진행하고 있음을 알 수 있다. 본 논문에서는 흉부와 복부 CT 영상에서 근감소증 진단을 위해서 흉추와 요추의 영역별 슬라이스를 검출하기 위해서 CNN 구조의 EfficientNetV2를 전이학습하여 인공지능 모듈을 개발하였다. 인공지능 모듈은 전체 흉부 및 복부 CT 영상에서 Cervical, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, L1, L2, L3, L4, L5, Sacral 총 19 클래스를 검출하도록 하였다. Test 데이터셋을 사용하여 Confusion Matrix와 Grad-CAM으로 모델의 정확도를 시각화하여 보였으며 검증으로 인공지능 모듈의 정확성을 측정하였다. 끝으로 우리가 개발한 다기관 공동연구 지원플랫폼에 적용하여 시각화된 결과를 보였다.

• Journal of KIBIM
• /
• v.14 no.2
• /
• pp.13-24
• /
• 2024

This research suggests a novel visualization approach utilizing Generative AI to render photorealistic architectural alternatives images in the early design phase. Photorealistic rendering intuitively describes alternatives and facilitates clear communication between stakeholders. Nevertheless, the conventional rendering process, utilizing 3D modelling and rendering engines, demands sophisticate model and processing time. In this context, the paper suggests a rendering approach employing the text-to-image method aimed at generating a broader range of intuitive and relevant reference images. Additionally, it employs an Text-to-Image method focused on producing a diverse array of alternatives reflecting architects' styles when visualizing the exteriors of residential buildings from the mass model images. To achieve this, fine-tuning for architects' styles was conducted using the Low-Rank Adaptation (LoRA) method. This approach, supported by fine-tuned models, allows not only single style-applied alternatives, but also the fusion of two or more styles to generate new alternatives. Using the proposed approach, we generated more than 15,000 meaningful images, with each image taking only about 5 seconds to produce. This demonstrates that the Generative AI-based visualization approach significantly reduces the labour and time required in conventional visualization processes, holding significant potential for transforming abstract ideas into tangible images, even in the early stages of design.