• Journal of Broadcast Engineering
• /
• v.26 no.2
• /
• pp.125-131
• /
• 2021

In recent years, Convolutional Neural Networks (CNNs) have achieved outstanding performance in the fields of computer vision such as image classification, object detection, visual quality enhancement, etc. However, as huge amount of computation and memory are required in CNN models, there is a limitation in the application of CNN to low-power environments such as mobile or IoT devices. Therefore, the need for neural network compression to reduce the model size while keeping the task performance as much as possible has been emerging. In this paper, we propose a method to compress CNN models by combining matrix decomposition methods of LR (Low-Rank) approximation and CP (Canonical Polyadic) decomposition. Unlike conventional methods that apply one matrix decomposition method to CNN models, we selectively apply two decomposition methods depending on the layer types of CNN to enhance the compression performance. To evaluate the performance of the proposed method, we use the models for image classification such as VGG-16, RestNet50 and MobileNetV2 models. The experimental results show that the proposed method gives improved classification performance at the same range of 1.5 to 12.1 times compression ratio than the existing method that applies only the LR approximation.

• The Journal of the Korea Contents Association
• /
• v.19 no.7
• /
• pp.1-12
• /
• 2019

According to a survey on the frequency of access to medical ads by the Korea Press Foundation in 2017, the most commonly exposed ads among adult men and women are advertising about beauty, plasticity and obesity. As of 2011, South Korea had the largest number of cosmetic surgeries in the world, with 131 cosmetic surgeries per 10,000 people. As a result, as many as 1,414 plastic surgery clinics are operating in South Korea, and the number is also on the rise. Although there are various standards for evaluating people's appearance, the desire to pursue a better look is growing day by day. Then, one might wonder what factors influence consumers' choices among the numerous advertisements for plastic surgery clinics. Based on these questions, this study identified the examples of plastic surgery advertisements, analyzed their type, and identified the types of advertisements with the high visual appeal of the advertising consumer through eye tracking experiment. In total, seven eye-tracking tests of plastic surgery social media advertisements were conducted on 10 subjects. The results showed that the commercial model was the biggest factor that caught the attraction and attention of the ad recipient first and that the most focused and long-standing factor was the treatment contents. Therefore, it is important to select proper commercial models for hospital and clinic contents and to specify factual treatment contents when producing social media advertisements for plastic surgeons. We hope these findings will help create online advertising for plastic surgery clinics effectively.

• Journal of Broadcast Engineering
• /
• v.23 no.6
• /
• pp.855-865
• /
• 2018

Sound event detection is one of the research areas to model human auditory cognitive characteristics by recognizing events in an environment with multiple acoustic events and determining the onset and offset time for each event. DCASE, a research group on acoustic scene classification and sound event detection, is proceeding challenges to encourage participation of researchers and to activate sound event detection research. However, the size of the dataset provided by the DCASE Challenge is relatively small compared to ImageNet, which is a representative dataset for visual object recognition, and there are not many open sources for the acoustic dataset. In this study, the sound events that can occur in indoor and outdoor are collected on a larger scale and annotated for dataset construction. Furthermore, to improve the performance of the sound event detection task, we developed a dual CNN structured sound event detection system by adding a supplementary neural network to a convolutional neural network to determine the presence of sound events. Finally, we conducted a comparative experiment with both baseline systems of the DCASE 2016 and 2017.

• Asian-Australasian Journal of Animal Sciences
• /
• v.32 no.7
• /
• pp.1015-1026
• /
• 2019

Objective: The objective of this study was to evaluate three different degrees of white striping (WS) addressing their automatic assessment and customer acceptance. The WS classification was performed based on a computer vision system (CVS), exploring different machine learning (ML) algorithms and the most important image features. Moreover, it was verified by consumer acceptance and purchase intent. Methods: The samples for image analysis were classified by trained specialists, according to severity degrees regarding visual and firmness aspects. Samples were obtained with a digital camera, and 25 features were extracted from these images. ML algorithms were applied aiming to induce a model capable of classifying the samples into three severity degrees. In addition, two sensory analyses were performed: 75 samples properly grilled were used for the first sensory test, and 9 photos for the second. All tests were performed using a 10-cm hybrid hedonic scale (acceptance test) and a 5-point scale (purchase intention). Results: The information gain metric ranked 13 attributes. However, just one type of image feature was not enough to describe the phenomenon. The classification models support vector machine, fuzzy-W, and random forest showed the best results with similar general accuracy (86.4%). The worst performance was obtained by multilayer perceptron (70.9%) with the high error rate in normal (NORM) sample predictions. The sensory analysis of acceptance verified that WS myopathy negatively affects the texture of the broiler breast fillets when grilled and the appearance attribute of the raw samples, which influenced the purchase intention scores of raw samples. Conclusion: The proposed system has proved to be adequate (fast and accurate) for the classification of WS samples. The sensory analysis of acceptance showed that WS myopathy negatively affects the tenderness of the broiler breast fillets when grilled, while the appearance attribute of the raw samples eventually influenced purchase intentions.

• Korean Journal of Anesthesiology
• /
• v.71 no.6
• /
• pp.467-475
• /
• 2018

Background: The Quality of Recovery-40 (QoR-40) is a widely-used, self-rated, and self-completed questionnaire for postoperative patients. The questionnaire is intended to elicit information from each patient regarding the quality of recovery during the postoperative period. It is noteworthy, however, that an official Korean version of the QoR-40 (QoR-40K) has not been established. The purpose of this study was to develop the QoR-40K by translation and cultural adaptation process and to evaluate the validity and reliability of the QoR-40K. Methods: After pre-authorization from the original author of the QoR-40, the translation procedure was established and carried out based upon Beaton's recommendation to create a QoR-40K model comparable to the original English QoR-40. Two hundred surgical patients were enrolled, and each completed the questionnaire during the preoperative period, on the third day, and 1 month after surgery. The QoR-40K was compared with the visual analogue scale (VAS) and another health-related questionnaire, the Short-form Health Survery-36 (SF-36). The method of validation for QoR-40K included test-retest reliability, internal consistency, and level of responsiveness. Results: Spearman's correlation coefficient for test-retest reliability was 0.895 (P < 0.001), and Cronbach's alpha of the global QoR-40K on the third day after surgery was 0.956. A positive correlation was obtained between the QoR-40K and the mental component summary of SF-36 (${\rho}=0.474$, P < 0.001), and a negative correlation was observed between QoR-40K and VAS (${\rho}=-0.341$, P < 0.001). The standardized responsive mean of the total QoR-40K was 0.71. Conclusions: The QoR-40K was found to be as acceptable and reliable as the original English QoR-40 for Korean patients after surgery, despite the apparent differences in the respective patients' cultural backgrounds.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.4
• /
• pp.577-584
• /
• 2021

Electrical energy is an essential component in present societies, which is an important basis for our technological society. In the design of new power infrastructure, it is important to consider the psychological aspects of how our culture considers and aspects its development as an integral component of the community environment. The construction of new high voltage overhead transmission lines has become a controversial issue for public policy of government due to social opposition. The members of community are concerned about how these power lines may have an impact on their lives, basically caused by their effects on health and safety. The landscape and visual impact is one of the most impact that can be easily perceived for local community. The computer 3D simulation of new landscape is illustrated by a real life use corresponding to the selection of the power line route with least observability for local community. This paper used ArcGIS(geographic information system tool) for planning, survey, basic route and detailed route, route for implementation of transmission line corridor. Also, the paper showed the map of natural environment, living environment, safety and altitude using database of power line corridor, and transmission siting model was developed by this study. The suggested landscape of computer simulation with lowest visibility on a power line zone can contribute to reducing oppositions of local community and accelerating the construction of new power lines.

• Journal of the Architectural Institute of Korea Planning & Design
• /
• v.34 no.9
• /
• pp.55-66
• /
• 2018

This study aims to propose color affective model of indoor space by evaluating subjects' physiological responses according to the types of behavioral and affective system. 99 subjects(44 females, 55 males) in their 20s, who had no disorders in visual perception, participated in the experiment. To categorize the subjects based on behavioral and affective system, BAS/BIS scale and Affective scale were used. Color stimuli were composed of five basic colors and three tones: vivid, pale and dull tone of R, Y, G, B, P. For physiological experiment, right and left prefrontal alpha activity was measured to analyze prefrontal EEG asymmetry. Participants were exposed to fifteen color stimuli for 20 seconds each other under the positive and negative emotional condition in a research room with the natural light blocked. The results and conclusion of this study are as follows. Along with factors of behavioral and affective system, cluster analysis was carried out and four types were classified. Type A had high BAS sensitivity, especially high 'drive' trait, and showed high levels of 'anxiety' and 'anger'. Type B had low BAS sensitivity, especially low 'fun seeking' and low 'drive' trait, and showed low levels of 'anxiety' as well as low levels of 'happiness'. Type C had low BIS sensitivity and showed high levels of 'happiness' and low levels of 'sadness'. Type D had high BIS sensitivity and showed high levels of 'lethargy' and 'sadness'. As a result of EEG signal analysis of color stimuli, Type B, Type C, and Type D showed significant differences in prefrontal alpha asymmetry under the negative emotional stimuli. Type B showed more left prefrontal activation in the spaces with pale R and dull G. Type C showed more left prefrontal activation in the spaces with vivid Y and B, pale R, and dull R, G, P. Type D showed more left prefrontal activation in the spaces with vivid Y and P, pale R, Y, P, and dull R, Y, G, B, P. The group of high BAS sensitivity was not influenced by color stimuli under the emotional conditions, whereas the group of high BIS sensitivity was affected by color stimuli under the negative emotional conditions. They showed left prefrontal activation when they were exposed the spaces with vivid, pale, dull tones of Y and P wall.

• Korean Journal of Otorhinolaryngology-Head and Neck Surgery
• /
• v.60 no.5
• /
• pp.209-214
• /
• 2017

Background and Objectives Tinnitus retraining therapy (TRT) is one of the most effective treatment modalities of tinnitus based on the neurophysiological model proposed by Jastreboff and Hazell. This study was performed to evaluate the effect of counselor factor on treatment outcomes of TRT. Subjects and Method The total of 78 patients who had TRT from three different counselors in a tinnitus clinic of tertiary referral center from Jan 2015 to Dec 2015 were included in this study. Their medical records were retrospectively reviewed to evaluate the therapeutic response to TRT. Results Among 78 patients who were followed-up for more than 6 months, 47, 20, and 11 patients were treated by counselors A, B, C (all ENT specialists), respectively. Counselor A had 15-year-experience of TRT counseling, whereas counselor B and C were well trained but beginners of TRT counseling. Initial clinical characteristics including Tinnitus Handicap Inventory (THI) and tinnitus Visual Analogue Scale (VAS) scores of the patients among three groups were not significantly different. Treatment responses evaluated via THI and most of the tinnitus VAS scores after at least 6 months after TRT were significantly improved in all three groups (p<0.05) with no significant difference between the senior (A) and junior (B, C) group. Conclusion TRT seems to be an effective treatment modality of tinnitus even in this short term follow-up study. Treatment outcomes of TRT may not depend on the counselors once they are well trained and follow the same protocol.

• The Journal of the Korea Contents Association
• /
• v.21 no.6
• /
• pp.104-112
• /
• 2021

In this study, emotional analysis was conducted based on the basic attribute data of music and the emotional model in psychology, and the result was applied to the visualization rules in the formative arts. In the existing studies using musical parameter, there were many cases with more practical purposes to classify, search, and recommend music for people. In this study, the focus was on enabling sound data to be used as a material for creating artworks and used for aesthetic expression. In order to study the music visualization as an art form, a method that can include human emotions should be designed, which is the characteristics of the arts itself. Therefore, a well-structured basic classification of musical attributes and a classification system on emotions were provided. Also, through the shape, color, and animation of the visual elements, the visualization of the musical elements was performed by reflecting the subdivided input parameters based on emotions. This study can be used as basic data for artists who explore a field of music visualization, and the analysis method and work results for matching emotion-based music components and visualizations will be the basis for automated visualization by artificial intelligence in the future.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.41 no.6
• /
• pp.751-760
• /
• 2021

Most bridges in Korea are over 20 years old, and many problems linked to their deterioration are being reported. The current practice for bridge inspection mainly depends on expert evaluation, which can be subjective. Recent studies have introduced data-driven methods using building information modeling, which can be more efficient and objective, but these methods require manual procedures that consume time and money. To overcome this, this study developed an image-based automaticbridge component classification network to reduce the time and cost required for converting the visual information of bridges to a digital model. The proposed method comprises two convolutional neural networks. The first network estimates the type of the bridge based on the superstructure, and the second network classifies the bridge components. In avalidation test, the proposed system automatically classified the components of 461 bridge images with 96.6 % of accuracy. The proposed approach is expected to contribute toward current bridge maintenance practice.