• Clinical Endoscopy
• /
• v.55 no.5
• /
• pp.594-604
• /
• 2022

Over the past decade, technological advances in deep learning have led to the introduction of artificial intelligence (AI) in medical imaging. The most commonly used structure in image recognition is the convolutional neural network, which mimics the action of the human visual cortex. The applications of AI in gastrointestinal endoscopy are diverse. Computer-aided diagnosis has achieved remarkable outcomes with recent improvements in machine-learning techniques and advances in computer performance. Despite some hurdles, the implementation of AI-assisted clinical practice is expected to aid endoscopists in real-time decision-making. In this summary, we reviewed state-of-the-art AI in the field of gastrointestinal endoscopy and offered a practical guide for building a learning image dataset for algorithm development.

• Journal of the Korea Institute of Building Construction
• /
• v.17 no.6
• /
• pp.545-557
• /
• 2017

The study is about the efficient alternative to concrete surface in the field of visual inspection technology for deteriorated infrastructure. By combining industrial drones and deep learning based image analysis techniques with traditional visual inspection and research, we tried to reduce manpowers, time requirements and costs, and to overcome the height and dome structures. On board device mounted on drones is consisting of a high resolution camera for detecting cracks of more than 0.3 mm, a lidar sensor and a embeded image processor module. It was mounted on an industrial drones, took sample images of damage from the site specimen through automatic flight navigation. In addition, the damege parts of the site specimen was used to measure not only the width and length of cracks but white rust also, and tried up compare them with the final image analysis detected results. Using the image analysis techniques, the damages of 54ea sample images were analyzed by the segmentation - feature extraction - decision making process, and extracted the analysis parameters using supervised mode of the deep learning platform. The image analysis of newly added non-supervised 60ea image samples was performed based on the extracted parameters. The result presented in 90.5 % of the damage detection rate.

• Journal of Broadcast Engineering
• /
• v.17 no.3
• /
• pp.519-528
• /
• 2012

Voice activity detection (VAD) is generally conducted by extracting features from the acoustic signal and a decision rule. The performance of such VAD algorithms driven by the input acoustic signal highly depends on the acoustic noise. When video signals are available as well, the performance of VAD can be enhanced by using the visual information which is not affected by the acoustic noise. Previous visual VAD algorithms usually use single visual feature to detect the lip activity, such as active appearance models, optical flow or intensity variation. Based on the analysis of the weakness of each feature, we propose to combine intensity change measure and the optical flow in the mouth region, which can compensate for each other's weakness. In order to minimize the computational complexity, we develop simple measures that avoid statistical estimation or modeling. Specifically, the optical flow is the averaged motion vector of some grid regions and the intensity variation is detected by simple thresholding. To extract the mouth region, we propose a simple algorithm which first detects two eyes and uses the profile of intensity to detect the center of mouth. Experiments show that the proposed combination of two simple measures show higher detection rates for the given false positive rate than the methods that use a single feature.

• Design Convergence Study
• /
• v.15 no.6
• /
• pp.195-214
• /
• 2016

Predictive analysis is based on a probabilistic learning algorithm called pattern recognition or machine learning. Therefore, if users want to extract more information from the data, they are required high statistical knowledge. In addition, it is difficult to find out data pattern and characteristics of the data. This study conducted statistical data analyses and visual data analyses to supplement prediction analysis's weakness. Through this study, we could find some implications that haven't been found in the previous studies. First, we could find data pattern when adjust data selection according as splitting criteria for the decision tree method. Second, we could find what type of data included in the final prediction model. We found some implications that haven't been found in the previous studies from the results of statistical and visual analyses. In statistical analysis we found relation among the multivariable and deducted prediction model to predict high box office performance. In visualization analysis we proposed visual analysis method with various interactive functions. Finally through this study we verified final prediction model and suggested analysis method extract variety of information from the data.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.17 no.12
• /
• pp.532-541
• /
• 2016

The objective of this paper is to support accurate pavement condition assessment and decision of proper maintenance method and time by conducting visual inspection and calculating the remaining life of pavement from falling weight deflectometer(FWD) data. Each was implemented in the same long-term performance pavement(LTPP) sections. Visual inspection was executed to measure pavement condition indices such as crack, rutting and international roughness index(IRI) and the Seoul Pavement Index(SPI) was calculated based on these results. The dynamic modulus was back-calculated from the FWD data. The remaining pavement lives were determined from equivalent single axle loading(ESAL) and FWD data. Correlation of maintenance priority by each result value was examined. Consequently, the correlation between remaining life to Crack and Rutting was higher than the other factors or indicesbecause IRI is not related to FWD value and SPI value consists with IRI value and other indices. The R-square value of correlation of FWD with Crack and Rutting was 0.65, which indicated an insufficient correlation. Consequently, when decision of maintenance of method, time, etc. is determined, FWD data have to be considered with Crack and Rutting because of those relations.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.2
• /
• pp.59-66
• /
• 2019

Small bridges carry out only general inspections based on visual inspection. The Bridges with more than 30 years of public use need to be decided on whether or not they will be reconstruction according to aging. However, there are some situations that need to be determined only by appearance condition, which is insufficient in terms of safety maintenance. In this paper, the condition evaluation and the load carrying capacity evaluation were carried out for aged small bridges. A comparison of the evaluation results was conducted to examine the appropriateness of the maintenance related to the decision making of the reconstruction. As a result of reviewing, two of the bridges showed that there are no abnormality in the safety of the state evaluation, but the load capacity were insufficient. Thus evaluation the safety and performing the reconstruction decision of aged small bridges by visual inspection alone with may cause problems. Therefore, it is necessary to carry out additional research on the ambient measurement and load carrying capacity evaluation for the maintenance of the bridges, and to supplement it through application of the bridge management system.

• Fashion & Textile Research Journal
• /
• v.7 no.4
• /
• pp.426-432
• /
• 2005

The health of feet is connected with individual's health and affects a man's activity. Shoes need to be designed to protect feet and to absorb the impact of land. Thus, design, comfort and economical efficiency are important factors of shoes. Consumers can choose suitable shoes for their feet in off-line shopping. However, in on-line shopping, because they can not wear shoes, compare to the off-line shopping, there are many problems in internet shopping. First, consumers can get limited information of shoes because they must search information of purchase without other's help. Second, because consumers can not get important information such as design, size and a comfort of wearing, they can not make a careful decision. Above these, the solution of user-oriented internet shopping is development of new type of prototype which is accessible to user and to offer visual information through 3D-virtual reality. The purpose of this study is to develpoment of shoes easy-order prototype so to reduce of internet shopping risk and offer to shopping convenience. Contents of the new prototype are as follows; 1. Consumers can print out a foot shaped plane figure and compare it to their foot. If the size of printed sheet is not proper, consumers can choose another size. Through this step, they can get their own size. 2. In 3D-virtual reality, consumers can get a various visual information. So they feel wearing comfort indirectly. It is expected that the result of this study would be an important case study to develop of shoes' easy-order prototype for a disabled person.

• Journal of Korea Water Resources Association
• /
• v.40 no.4
• /
• pp.313-324
• /
• 2007

An irregular cell-based numerical model was developed to analyze underground space flooding. In this model, the flow characteristics in underground space were computed by link-node system. Also, the model can simulate the underground flood flow related to the influence of stairs and wall-structures. Empirical discharge formula were introduced to analyze weir-type flow for shopping mall, and channel-type flow for subway railroad respectively. The simulated results matched in reasonable range compared with the observed depth. The dual-drainage inundation analysis model and the underground space flood analysis model were integrated using visual basic application of ArcGIS system. The developed model can help the decision support system of flood control authority for redesigning and constructing flood prevention structures and making the potential inundation zone, and establishing flood-mitigation measures.

• Journal of Environmental Impact Assessment
• /
• v.8 no.4
• /
• pp.47-57
• /
• 1999

This study is aimed at the development of comprehensive environmental management system, which can be operated on the basis of world wide web, as a topic of G7 project. Even though there should be lots of works remaining to achieve this goal, preliminary products can be summarized as follows : 1) integrated environmental information management system, 2) web based control engine, 3) surface water environment management system, 4) subsurface water environment management system, 5) sewer and waterworks management system. The core methodology of the engine is the generalized multidimensional finite element matrices to depict the terms in the analysis of various partial differential equations. Spatial information management system (ArcView) and Visual Basic were extensively employed to construct GUI oriented web based engine. The developed systems were composed of very intense computer codes due to the necessity of combinatory management of environmental problems. The web based engine could be served as a decision tool for the integrated management of environmental projects in Korea.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.35S no.8
• /
• pp.86-95
• /
• 1998

Intra-frame video compression, which cannot make use of temporal predictions, requires much higher bit rates compared with inter-frame schemes. In order to reduce bit rates, intra-frame predictive coding methods in DCT domain have been studied especially within the framework of the MPEG-4 video coding standard currently being developed. In this paper, we propose novel intra-frame predictive coding methods in DCT domain with the marginal complexity increase over the conventional methods . The proposed methods consist of a DC coefficient prediction method and two AC coefficient prediction methods. The proposed methods consist of a DC coefficient prediction method and two AC coefficient prediction methods. The proposed DC coefficient prediction method makes it possible to adaptively select the prediction directions without overhead bits, by comparing gradients of DC coefficients from neighboring blocks. As the AC coefficient prediction methods, first, we present an effective method which can improve the prediction directions of the MPEG-4 scheme by considering the DC coefficient of the current block to be coded. And, we present another effective method that decision on the prediction is carried out for each AC coefficient. Simulation results show that substantial bit savings can be achieved by the proposed methods.