• Journal of the Korea Society of Computer and Information
• /
• v.24 no.11
• /
• pp.29-39
• /
• 2019

As mobile robot applications increase in real life, the need of low cost autonomous driving are gradually increasing. We propose a novel vision-based real-time lane detection and tracking system that supports autonomous driving of mobile robots in constrained tracks which are designed considering indoor driving conditions of mobile robots. Considering the processing of lanes with various shapes and the pre-adjustment of operation parameters, the system structure with multi-operation modes are designed. In parameter tuning mode, thresholds of the color filter is dynamically adjusted based on the geometric property of the lane thickness. And in the unstable input mode of curved tracks and the stable input mode of straight tracks, lane feature pixels are adaptively extracted based on the geometric and temporal characteristics of the lanes and the lane model is fitted using the least-squared method. The track centerline is calculated using lane models and the motion model is simplified and tracked by a linear Kalman filter. In the driving experiments, it was confirmed that even in low-performance robot configurations, real-time processing produces the accurate autonomous driving in the constrained track.

• International Journal of Computer Science & Network Security
• /
• v.21 no.1
• /
• pp.27-33
• /
• 2021

The whole world now is dealing with Coronavirus, and it has turned to be one of the most widespread and long-lived pandemics of our times. Reports reveal that the infectious disease has taken toll of the almost 80% of the world's population. Amidst a lot of research going on with regards to the prediction on growth and transmission through Symptomatic carriers of the virus, it can't be ignored that pre-symptomatic and asymptomatic carriers also play a crucial role in spreading the reach of the virus. Classification Algorithm has been widely used to classify different types of COVID-19 carriers ranging from simple feature-based classification to Convolutional Neural Networks (CNNs). This research paper aims to present a novel technique using a Random Forest Machine learning algorithm with hyper-parameter tuning to classify different types COVID-19-carriers such that these carriers can be accurately characterized and hence dealt timely to contain the spread of the virus. The main idea for selecting Random Forest is that it works on the powerful concept of "the wisdom of crowd" which produces ensemble prediction. The results are quite convincing and the model records an accuracy score of 99.72 %. The results have been compared with the same dataset being subjected to K-Nearest Neighbour, logistic regression, support vector machine (SVM), and Decision Tree algorithms where the accuracy score has been recorded as 78.58%, 70.11%, 70.385,99% respectively, thus establishing the concreteness and suitability of our approach.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.1
• /
• pp.44-49
• /
• 2021

Recently, due to COVID-19, studies have been popularly worked to apply neural network to mask wearing automatic detection system. For applying neural networks, the 1-stage detection or 2-stage detection methods are used, and if data are not sufficiently collected, the pretrained neural network models are studied by applying fine-tuning techniques. In this paper, the system is consisted of 2-stage detection method that contain MTCNN model for face recognition and ResNet model for mask detection. The mask detector was experimented by applying five ResNet models to improve accuracy and fps in various environments. Training data used 17,217 images that collected using web crawler, and for inference, we used 1,913 images and two one-minute videos respectively. The experiment showed a high accuracy of 96.39% for images and 92.98% for video, and the speed of inference for video was 10.78fps.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.16 no.6
• /
• pp.369-377
• /
• 2023

In the industrial manufacturing sector, quality control is pivotal for minimizing defect rates; inadequate management can result in additional costs and production delays. This study underscores the significance of detecting texture defects in manufactured goods and proposes a more precise defect detection technique. While the DFR(Deep Feature Reconstruction) model adopted an approach based on feature map amalgamation and reconstruction, it had inherent limitations. Consequently, we incorporated a new loss function using statistical methodologies, integrated a skip connection structure, and conducted parameter tuning to overcome constraints. When this enhanced model was applied to the texture category of the MVTec-AD dataset, it recorded a 2.3% higher Defect Segmentation AUC compared to previous methods, and the overall defect detection performance was improved. These findings attest to the significant contribution of the proposed method in defect detection through the reconstruction of feature map combinations.

• Journal of Powder Materials
• /
• v.31 no.2
• /
• pp.119-125
• /
• 2024

The n-type Bi_2-xSb_xTe₃ compounds have been of great interest due to its potential to achieve a high thermoelectric performance, comparable to that of p-type Bi_2-xSb_xTe₃. However, a comprehensive understanding on the thermoelectric properties remains lacking. Here, we investigate the thermoelectric transport properties and band characteristics of n-type Bi_2-xSb_xTe₃ (x = 0.1 - 1.1) based on experimental and theoretical considerations. We find that the higher power factor at lower Sb content results from the optimized balance between the density of state effective mass and nondegenerate mobility. Additionally, a higher carrier concentration at lower x suppresses bipolar conduction, thereby reducing thermal conductivity at elevated temperatures. Consequently, the highest zT of ~ 0.5 is observed at 450 K for x = 0.1 and, according to the single parabolic band model, it could be further improved by ~70 % through carrier concentration tuning.

• Convergence Security Journal
• /
• v.24 no.2
• /
• pp.201-207
• /
• 2024

Voice is a critical element of human communication, and the development of speech recognition models is one of the significant achievements in artificial intelligence, which has recently been applied in various aspects of human life. The application of speech recognition models in the military field is also inevitable. However, before artificial intelligence models can be applied in the military, it is necessary to research their vulnerabilities. In this study, we evaluates the military applicability of the multilingual speech recognition model "Whisper" by examining its vulnerabilities to battlefield noise, white noise, and adversarial attacks. In experiments involving battlefield noise, Whisper showed significant performance degradation with an average Character Error Rate (CER) of 72.4%, indicating difficulties in military applications. In experiments with white noise, Whisper was robust to low-intensity noise but showed performance degradation under high-intensity noise. Adversarial attack experiments revealed vulnerabilities at specific epsilon values. Therefore, the Whisper model requires improvements through fine-tuning, adversarial training, and other methods.

• 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 Korea Academia-Industrial cooperation Society
• /
• v.21 no.10
• /
• pp.8-15
• /
• 2020

A forecasting method using deep learning does not have consistent results due to the differences in the characteristics of the dataset, even though they have the same forecasting models and parameters. For example, the forecasting model X optimized with dataset A would not produce the optimized result with another dataset B. The forecasting model with the characteristics of the dataset needs to be optimized to increase the accuracy of the forecasting model. Therefore, this paper proposes novel optimization steps for outlier removal, dataset classification, and a CNN-LSTM-based hyperparameter tuning process to forecast the daily power usage of a university campus based on the hourly interval. The proposing model produces high forecasting accuracy with a 2% of MAPE with a single power input variable. The proposing model can be used in EMS to suggest improved strategies to users and consequently to improve the power efficiency.

• Journal of the Society of Naval Architects of Korea
• /
• v.55 no.5
• /
• pp.401-414
• /
• 2018

Planing hull type ships are often equipped with interceptor or trim tab to improve the excessive trim angle which leads to poor resistance and sea keeping performances. The purpose of this study is to design a controller to control the attitude of the ship by controllable stern interceptor and validate the effectiveness of the attitude control by the towing tank test. Embedded controller, servo motor and controllable stern interceptor system were equipped with planing hull type model ship. Prior to designing the control algorithm, a model test was performed to identify the system dynamic model of the planing hull type ship including the stern interceptor. The matrix components of model were optimized by Genetic Algorithm. Using the identified model, PID controller which is a classical controller and sliding mode controller which is a nonlinear robust controller were designed. Gain tuning of the controllers and running simulation was conducted before the towing tank test. Inserting the designed control algorithm into the embedded controller of the model ship, the effectiveness of the active control of the stern interceptor was validated by towing tank test. In still water test with small disturbance, the sliding mode controller showed better performance of canceling the disturbance and the steady-state control performance than the PID controller.

• Asian-Australasian Journal of Animal Sciences
• /
• v.32 no.12
• /
• pp.1836-1843
• /
• 2019

Objective: Social genetic effects (SGE) are an important genetic component for growth, group productivity, and welfare in pigs. The present study was conducted to evaluate i) the feasibility of the single-step genomic best linear unbiased prediction (ssGBLUP) approach with the inclusion of SGE in the model in pigs, and ii) the changes in the contribution of heritable SGE to the phenotypic variance with different scaling ${\omega}$ constants for genomic relationships. Methods: The dataset included performance tested growth rate records (average daily gain) from 13,166 and 21,762 pigs Landrace (LR) and Yorkshire (YS), respectively. A total of 1,041 (LR) and 964 (YS) pigs were genotyped using the Illumina PorcineSNP60 v2 BeadChip panel. With the BLUPF90 software package, genetic parameters were estimated using a modified animal model for competitive traits. Giving a fixed weight to pedigree relationships (${\tau}:1$), several weights (${\omega}_{xx}$, 0.1 to 1.0; with a 0.1 interval) were scaled with the genomic relationship for best model fit with Akaike information criterion (AIC). Results: The genetic variances and total heritability estimates ($T^2$) were mostly higher with ssGBLUP than in the pedigree-based analysis. The model AIC value increased with any level of ${\omega}$ other than 0.6 and 0.5 in LR and YS, respectively, indicating the worse fit of those models. The theoretical accuracies of direct and social breeding value were increased by decreasing ${\omega}$ in both breeds, indicating the better accuracy of ${\omega}_{0.1}$ models. Therefore, the optimal values of ${\omega}$ to minimize AIC and to increase theoretical accuracy were 0.6 in LR and 0.5 in YS. Conclusion: In conclusion, single-step ssGBLUP model fitting SGE showed significant improvement in accuracy compared with the pedigree-based analysis method; therefore, it could be implemented in a pig population for genomic selection based on SGE, especially in South Korean populations, with appropriate further adjustment of tuning parameters for relationship matrices.