• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.18 no.3
• /
• pp.106-118
• /
• 2019

In recent years, research and development on image object recognition using artificial intelligence have been actively carried out, and it is expected to be used for road maintenance. Among them, artificial intelligence models for object detection of road surface are continuously introduced. In order to develop such object recognition algorithms, a backbone network that extracts feature maps is essential. In this paper, we will discuss how to select the appropriate neural network. To accomplish it, we compared with 4 different deep neural networks using 6,000 road surface damage images. Based on three evaluation methods for analyzing characteristics of neural networks, we propose a method to determine optimal neural networks. In addition, we improved the performance through optimal tuning of hyper-parameters, and finally developed a light backbone network that can achieve 85.9% accuracy of road surface damage classification.

• The Journal of the Korea Contents Association
• /
• v.19 no.3
• /
• pp.72-79
• /
• 2019

Autonomous locomotion has two essential functionalities: mapping builds and updates maps by uncertain position information and measured sensor inputs, and localization is to find the positional information with the inaccurate map and the sensor information. In addition, obstacle detection, avoidance, and path designs are necessarily required for autonomous locomotion by combining the probabilistic methods based on uncertain locations. The sensory inputs, which are measured by a metric-based scanner, have difficulties of distinguishing moving obstacles like humans from static objects like walls in given environments. This paper proposes the low resolution grid map combined with reinforcement learning, which is compared with the conventional recognition method for detecting static and moving objects to generate obstacle avoiding path. Finally, the proposed method is verified with experimental results.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.3
• /
• pp.157-165
• /
• 2021

Recently, as the spatial resolution of satellite and aerial images has improved, various studies using remotely sensed data with high spatial resolution have been conducted. In particular, since the building extraction is essential for creating digital thematic maps, high accuracy of building extraction result is required. In this manuscript, building extraction models were generated using SegNet, U-Net, FC-DenseNet, and HRNetV2, which are representative semantic segmentation models in deep learning techniques, and then the evaluation of building extraction results was performed. Training dataset for building extraction were generated by using aerial orthophotos including various buildings, and evaluation was conducted in three areas. First, the model performance was evaluated through the region adjacent to the training dataset. In addition, the applicability of the model was evaluated through the region different from the training dataset. As a result, the f1-score of HRNetV2 represented the best values in terms of model performance and applicability. Through this study, the possibility of creating and modifying the building layer in the digital map was confirmed.

• Korean Medical Education Review
• /
• v.24 no.2
• /
• pp.128-140
• /
• 2022

The purpose of this study was to redesign a problem-based learning (PBL) curriculum and compare the differences between the previous and redesigned PBL based on the results of course satisfaction and student assessments. The PBL was redesigned using curriculum design guidelines (including revisions of curriculum objectives, learning components, learning environments, and assessment methods) that were developed based on previous studies and evaluation results. A comparative study was employed using course satisfaction surveys from the previous and redesigned curricula, and a total of 45 students participated. We also compared student assessment results from concept mapping, learning issue reports, modified essay questions, and reflection journals. We identified four key findings. First, we explored the possibility that the redesigned PBL could be implemented by student facilitators without professors as tutors. Second, the redesigned PBL fostered group dynamics that facilitated developing communication skills and collaborative learning through small-group discussions. Third, the new learning elements added in the redesigned PBL made a meaningful contribution to enhancing students' clinical reasoning based on hypothetico-deductive reasoning. Fourth, concept maps in redesigned PBL contained more complex and various nodes and connections, and the levels of the nodes were more appropriate. The implications of this study can provide meaningful preliminary information for redesigning PBL curricula for medical students to develop their essential competencies through PBL.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2021.06a
• /
• pp.152-152
• /
• 2021

The importance of groundwater has long been recognized, but the ground water potential to become contaminated as a result of human activities has only been recognized in recently. Before 1980 it was thought that soils served as filters, preventing harmful substances deposited at the surface from migrating into groundwater. Today it is known that soils have a finite capacity to protect groundwater. It can be contaminated from divers sources. Therefore, Assessment of aquifer vulnerability to pollution is essential for the protection and management of groundwater and land use planning. In this study, we used DRASTIC and AVI for groundwater vulnerability to contamination assessment. the different methods were applied to the southern coastal sedimentary basin of Benin and DRASTIC method was modified in two different steps. First, we modified DRASTIC by adding land use parameter to include the actual pollution sources (DRASTIC_LcLu) and second, classic DRASTIC weights was modified using Shannon's entropy (Entropy weight DRASTIC). The reliability of the applied approaches was verified using nitrate (NO3^-) concentration and by comparing the overall vulnerability maps to the previous researches in the study area and in the world. The results from validation showed that the addition of landcover/land use parameter to the classic DRASTIC helps to improve the method for better definition of the vulnerable areas in the basin and also, the weight modification using entropy improved better the method because Entropy weight DRASTIC_LcLu showed the highest correlation with nitrate concentration in the study basin. In summary the weight modification using entropy approach reduced the uncertainty of the human subjectivity in assigning weights and ratings in the standard DRASTIC.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2020.06a
• /
• pp.172-172
• /
• 2020

In recent years, soil erosion has come to be regarded as an essential environmental problem in human life. Soil erosion causes various on- and off-site problems such as ecosystem destruction, decreased agricultural productivity, increased riverbed deposition, and deterioration of water quality in streams. To solve these problems caused by soil erosion, it is necessary to quantify where, when, how much soil erosion occurs. Empirical erosion models such as the Universal Soil Loss Equation (USLE) family models have been widely used to make spatially distributed soil erosion vulnerability maps. Even if the models detect vulnerable sites relatively well by utilizing big data related to climate, geography, geology, land use, etc. within study domains, they do not adequately describe the physical process of soil erosion on the ground surface caused by rainfall or overland flow. In other words, such models remain powerful tools to distinguish erosion-prone areas at the macro scale but physics-based models are necessary to better analyze soil erosion and deposition and eroded particle transport. In this study, the physics-based Surface Soil Erosion Model (SSEM) was upgraded based on field survey information to produce sediment yield at the watershed scale. The modified model (hereafter MoSE) adopted new algorithms on rainfall kinematic energy and surface flow transport capacity to simulate soil erosion more reliably. For model validation, we applied the model to the Doam dam watershed in Gangwon-do and compared the simulation results with the USLE outputs. The results showed that the revised physics-based soil erosion model provided more improved and reliable simulation results than the USLE in terms of the spatial distribution of soil erosion and deposition.

• Journal of Advanced Navigation Technology
• /
• v.26 no.6
• /
• pp.474-480
• /
• 2022

Optical phase conjugation is one of techniques capable of compensating for distortion due to chromatic dispersion and nonlinearity, which are essential for long-distance transmission of wavelength division multiplexed (WDM) signal. We proposed and analyzed a way to solve the limitations of this technology through dispersion map with periodic dispersion profile. In the proposed system, optical phase conjugator (OPC) is placed at the position of 1:2 or 2:1 of the entire link, and the dispersion profile of dispersion map has periodic shape in the form of a sine wave or an inverse-sine wave. It was confirmed that the effective compensation of the distorted 960 Gb/s WDM signal was further improved through the proposed periodic dispersion map when the OPC was located at the 1:2 point instead of the 2:1 point of the entire link. In addition, it was found that the maximum RDPS allocated to fiber span should be 1,800 ps/nm or more in order to increase the design flexibility of dispersion-managed link with the proposed periodic dispersion map.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.11
• /
• pp.2924-2944
• /
• 2023

Alzheimer's disease (AD) is a neurological condition that is recognized as one of the primary causes of memory loss. AD currently has no cure. Therefore, the need to develop an efficient model with high precision for timely detection of the disease is very essential. When AD is detected early, treatment would be most likely successful. The most often utilized indicators for AD identification are the Mini-mental state examination (MMSE), and the clinical dementia. However, the use of these indicators as ground truth marking could be imprecise for AD detection. Researchers have proposed several computer-aided frameworks and lately, the supervised model is mostly used. In this study, we propose a novel 3D Convolutional Neural Network Multilayer Perceptron (3D CNN-MLP) based model for AD classification. The model uses Attention Mechanism to automatically extract relevant features from Magnetic Resonance Images (MRI) to generate probability maps which serves as input for the MLP classifier. Three MRI scan categories were considered, thus AD dementia patients, Mild Cognitive Impairment patients (MCI), and Normal Control (NC) or healthy patients. The performance of the model is assessed by comparing basic CNN, VGG16, DenseNet models, and other state of the art works. The models were adjusted to fit the 3D images before the comparison was done. Our model exhibited excellent classification performance, with an accuracy of 91.27% for AD and NC, 80.85% for MCI and NC, and 87.34% for AD and MCI.

• Journal of the Korean Geographical Society
• /
• v.40 no.3 s.108
• /
• pp.296-309
• /
• 2005

In this study, we investigated the change of future land-surface and relationships of land-surface change with geo-spatial information, using a Bayesian prediction model based on a likelihood ratio function, for analysing the land-surface change of the Gongju area. We classified the land-surface satellite images, and then extracted the changing area using a way of post classification comparison. land-surface information related to the land-surface change is constructed in a GIS environment, and the map of land-surface change prediction is made using the likelihood ratio function. As the results of this study, the thematic maps which definitely influence land-surface change of rural or urban areas are elevation, water system, population density, roads, population moving, the number of establishments, land price, etc. Also, thematic maps which definitely influence the land-surface change of forests areas are elevation, slope, population density, population moving, land price, etc. As a result of land-surface change analysis, center proliferation of old and new downtown is composed near Gum-river, and the downtown area will spread around the local roads and interchange areas in the urban area. In case of agricultural areas, a small tributary of Gum-river or an area of local roads which are attached with adjacent areas showed the high probability of change. Most of the forest areas are located in southeast and from this result we can guess why the wide chestnut-tree cultivation complex is located in these areas and the capability of forest damage is very high. As a result of validation using a prediction rate curve, a capability of prediction of urban area is $80\%$, agriculture area is $55\%$, forest area is $40\%$ in higher $10\%$ of possibility which the land-surface change would occur. This integration model is unsatisfactory to Predict the forest area in the study area and thus as a future work, it is necessary to apply new thematic maps or prediction models In conclusion, we can expect that this way can be one of the most essential land-surface change studies in a few years.

• Korean Journal of Soil Science and Fertilizer
• /
• v.42 no.1
• /
• pp.21-28
• /
• 2009

Detailed information on soil characteristics is of great importance for the use and conservation of soil resources that are essential for human welfare and ecosystem sustainability. This paper introduces soil inventory of Korea focusing on national soil database establishment, information systems, use, and future direction for natural resources management. Different scales of soil maps surveyed and soil test data collected by RDA (Rural Development Administration) were computerized to construct digital soil maps and database. Soil chemical properties and heavy metal concentrations in agricultural soils including vulnerable agricultural soils were investigated regularly at fixed sampling points. Internet-based information systems for soil and agro-environmental resources were developed based on 'National Soil Survey Projects' for managing soil resources and for providing soil information to the public, and 'Agroenvironmental Change Monitoring Project' to monitor spatial and temporal changes of agricultural environment will be opened soon. Soils data has a great potential of further application in estimation of soil carbon storage, water capacity, and soil loss. Digital mapping of soil and environment using state-of-the-art and emerging technologies with a pedometrics concept will lead to future direction.