• Convergence Security Journal
• /
• v.21 no.2
• /
• pp.57-66
• /
• 2021

Deep neural networks (DNNs) provide excellent performance for image, speech, and pattern recognition. However, DNNs sometimes misrecognize certain adversarial examples. An adversarial example is a sample that adds optimized noise to the original data, which makes the DNN erroneously misclassified, although there is nothing wrong with the human eye. Therefore studies on defense against adversarial example attacks are required. In this paper, we have experimentally analyzed the success rate of detection for adversarial examples by adjusting various parameters. The performance of the ensemble defense method was analyzed using fast gradient sign method, DeepFool method, Carlini & Wanger method, which are adversarial example attack methods. Moreover, we used MNIST as experimental data and Tensorflow as a machine learning library. As an experimental method, we carried out performance analysis based on three adversarial example attack methods, threshold, number of models, and random noise. As a result, when there were 7 models and a threshold of 1, the detection rate for adversarial example is 98.3%, and the accuracy of 99.2% of the original sample is maintained.

• Journal of IKEEE
• /
• v.27 no.3
• /
• pp.340-344
• /
• 2023

In this paper, we propose a study on the development of deep learning structure to secure visibility of outdoor LED display board according to weather change. The proposed technique secures the visibility of the outdoor LED display board by automatically adjusting the LED luminance according to the weather change using deep learning using an imaging device. In order to automatically adjust the LED luminance according to weather changes, a deep learning model that can classify the weather is created by learning it using a convolutional network after first going through a preprocessing process for the flattened background part image data. The applied deep learning network reduces the difference between the input value and the output value using the Residual learning function, inducing learning while taking the characteristics of the initial input value. Next, by using a controller that recognizes the weather and adjusts the luminance of the outdoor LED display board according to the weather change, the luminance is changed so that the luminance increases when the surrounding environment becomes bright, so that it can be seen clearly. In addition, when the surrounding environment becomes dark, the visibility is reduced due to scattering of light, so the brightness of the electronic display board is lowered so that it can be seen clearly. By applying the method proposed in this paper, the result of the certified measurement test of the luminance measurement according to the weather change of the LED sign board confirmed that the visibility of the outdoor LED sign board was secured according to the weather change.

• Journal of the Korean Institute of Rural Architecture
• /
• v.14 no.4
• /
• pp.107-116
• /
• 2012

This study aims to investigate what degree do the hospital buildings carry out their role in regional scenery. To carry out this study the existing theoretical and field survey was conducted to the 44 private hospitals in small and medium-sized cities - Jeonju city, Kunsan city and Iksan city in Jeonlabukdo district. The contents of the survey are the general aspect of the polled, the image of the buildings, that of medical treatment parts, the relations to the regional scenery. And they are questionnaire surveyed to the 44 private hospital buildings. In questionnaire survey, the 35 professional architects who major architecture and related design and 118 citizens who doesn't major architecture are participated. The results of this study are as follows: 1. Those surveyed who answered those private hospital buildings which have their functional images are barely 20%. Some of 27% answered they have images of the commercial building. Especially in professional architects, 5% of them answered thet have their hospital images. It showed that they don't have their identity in the architectural expression in the processof architectural design. 2. The architectural facade elements that express their images in surveyed buildings are answered that architectural building form is 62%, architectural color is 27% and architectural material is 10% in turn. And the architectural elements that express their images in surveyed buildings are answered that building wall is 56%, architectural decoration is 32%. In this article, the reason why the ratio of the architectural decoration is high is that they don' have their identity in building itself. And they expressed their identity by their decoration, sign system not their buildings. 3. As the preference level for their images, some of the 49% of the total surveyed answered common. The general surveyed answered a little preference. But the architect professional group answered preference is 14%, not preference is 29%. The reason of this article is that the general surveyed are familiar with their buildings. So the questionnaire of the harmony with the surrounding scenery, 25% of the general surveyed answered that they are in harmony with surrounding but 22% of them answered that they are not in them. But In professional group, they answered 6% oh them are positive 37% of them are negative. This results suggests that the identity of the private hospital buildings are established in the process of archiectural design.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.2
• /
• pp.31-41
• /
• 2024

Deep learning models show excellent performance in tasks such as image classification and object detection in the field of computer vision, and are used in various ways in actual industrial sites. Recently, research on improving robustness has been actively conducted, along with pointing out that this deep learning model is vulnerable to hostile examples. A hostile example is an image in which small noise is added to induce misclassification, and can pose a significant threat when applying a deep learning model to a real environment. In this paper, we tried to confirm the robustness of the edge-learning classification model and the performance of the adversarial example detection model using it for adversarial examples of various algorithms. As a result of robustness experiments, the basic classification model showed about 17% accuracy for the FGSM algorithm, while the edge-learning models maintained accuracy in the 60-70% range, and the basic classification model showed accuracy in the 0-1% range for the PGD/DeepFool/CW algorithm, while the edge-learning models maintained accuracy in 80-90%. As a result of the adversarial example detection experiment, a high detection rate of 91-95% was confirmed for all algorithms of FGSM/PGD/DeepFool/CW. By presenting the possibility of defending against various hostile algorithms through this study, it is expected to improve the safety and reliability of deep learning models in various industries using computer vision.

• Archives of design research
• /
• v.18 no.4 s.62
• /
• pp.287-296
• /
• 2005

The street image in a business area is so much affected by Facade that the front side of a roadside building makes. Recently, for the indiscreet and intemperate advertising signboard of the front side of roadside buildings, a streetscape becomes more disordered than before, so now we need to do research about signboards of roadside buildings for a streetscape image. In this research, we focused on a streetscape with difference of occupation density of signboard in the business area via investigation and analysis about occupation density of signboards of the front side of roadside buildings, and we suggested optimum occupation density of signboards for supporting the road image positively. An object of research is the street in the business area that has many pedestrians and active passing zone of cars. We investigated and analyzed how to feel street images on the rate of occupation density of roadside building's signboards of in the chosen street. As a result of using an adjective that we use for estimating street view images for extraction of street images, we could know 2 factors. We named that one is the image of recognition, and the other is the image of feelings. We knew that signboard density of street of heavily recognized images is from 20% to 30% and, signboard density of street of heavily feeling images is from 50% to 60%. We also could know that people feel both images of recognition and images of feeling in specific density, 30 to 50%. Through this result of research, we can suggest Facade on signboard density with the recognition and the feeling and use images of the street view as materials to be more specific and more special.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.14 no.4
• /
• pp.13-21
• /
• 2011

Update of road facilities database such as road sign, traffic lights, and street lights is interesting business in a local government. Recently, existing road facilities database, aerial photo and topographic map are referred for the installation and complement of road facilities. But it is difficult to comprehend road facilities' condition and additional expenses may appear in field survey. Therefore, it is necessary to establish and update road facility DB and many studies has been carried out to efficiently collect road related spatial data. In this study, the establishment of various complicated road facility DB was conducted by images that had been obtained by digital camera with a built-in bluetooth and DGPS. Results showed that road facility DB was constructed effectively and suggested the possibility of road facility management using images based on coordinate through accuracy analyses using total-station surveying. And using digital camera and DGPS is expected to effective real-time update and management of road facility DB.

• Journal of the Korean Chemical Society
• /
• v.52 no.3
• /
• pp.315-321
• /
• 2008

The posters are the tools to communicate authors' idea with others by visual image and a little word. Every discipline has its own sign with specific meaning shared by the members of the discipline. Chemists and students learning chemistry, therefore, will communicate with each other by specific chemical signs (codes). The Korean Chemical Society has held the feast of drawing chemistry posters by students nationwide since 2004. In 2004 and 2005, more than three thousands of posters were submitted, and about one hundred and fifty posters were selected as the prize winners. The award was divided by the grade levels of elementary, middle, and high school. This study explores the codes of chemistry used in students' posters. With the analysis of the visual elements and the verbal elements of posters, 7 chemical codes were found such as the liquid, the experiment apparatus, the graduations, the chemical language, scientists, the earth and environment, and things around us. In addition, the differences were investigated on the grade levels, on awarded or non-awarded poster, and on the years. Educational implications of these findings are discussed.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.5
• /
• pp.617-623
• /
• 2008

This paper presents a human motion recognition method using both centroid shift and local basis images. The centroid shift based on 1st moment balance technique is applied to get the robust motion images against position or size changes, the extraction of local basis images based on independent component analysis(ICA) is also applied to find a set of statistically independent motion features, which is included in each motions. Especially, ICA of fixed-point(FP) algorithm based on Newton method is used for being quick to extract a local basis images of motions. The proposed method has been applied to the problem for recognizing the 160(1 person * 10 animals * 16 motions) sign language motion images of 240*215 pixels. The 3 distances such as city-block, Euclidean, negative angle are used as measures when match the probe images to the nearest gallery images. The experimental results show that the proposed method has a superior recognition performances(speed, rate) than the method using local eigen images and the method using local basis images without centroid shift respectively.

• Clinics in Shoulder and Elbow
• /
• v.2 no.1
• /
• pp.35-40
• /
• 1999

Impingement syndrome is caused by a conflictual status between rotator cuff, subacromial bursa and anatomic and functional coracoacromial arch. The purpose of this study was to assessment the coracoacromial arch by MRI and to determine major factors among five components of coracoacromial arch. We analyzed forty-two cases of clinical impingement sign and test positive and postoperative confirmed diagnosed from March, 1991 to January, 1999. We evaluated acromial end abnormality according to the Bigliani acromial type and formation of osteophyte. Clavicular end abnormality classified flat, outward protrusion, inward protrusion to coracoacromial arch. Acromioclavicular joint abnormalities were advanced osteoarthritis and positive signal change. Coracoacromial ligament thickening was above 2 mm in oblique sagittal image. Coracoid process abnormality was inward protrusion to coracoacromial arch. All consecutive patients abnormalities were as follows: clavicular end osteophyte formation and inward protrusion to coracoacrmial arch were 30%, acromial end osteophyte formation was 28%, advanced acromioclavicular joint arthritis and osteophyte formation were 56%, coracoacromial ligament thickening was 24% and no coracoid process inward protrusion to coracoacromial arch. Impingement syndrome combined with rotator cuff tear group abnormalities were clavicular end(40%), acromial end(40%), acromioclavicular joint(20%), coracoacromialligament(20%) and coracoid process abnormality(0%) respectively. Only impingement syndrome group abnormalities were clavicular end(25%), acromial end(31%), acromioclavicular joint(62%), coracoacromial ligament(25%) and coracoid process(0%) respectively. Acromial type I(flat) were 6 cases, type II(curved) were 26 cases and type III(hooked) were 10 cases. We concluded that the most important contributing factors for impingement syndrome was acromial type and second was acromioclavicular joint arthritis and bony spur formation.

• Journal of the Korean Society of Costume
• /
• v.35
• /
• pp.181-193
• /
• 1997

The objectives of this study were to classify bedclothes purchase motives and to examine how bedclothes purchase mo-tives and design preferences very accord-ing to age and income. Samples were 217 housewives residing in metropolitan area. Questionnaire included 17 Likert type items of bedclothes pur-chase motive measure 12 items of color preference 7 items of pattern (floral geo-metric abstract stripe plaid polka dot, and paisley pattern). The results of the study were as fol-lows. 1. 5 factors of bedclothes purchase mo-tives were derived by factor analysis ; F .1 'design': F 2. "brand' ; F.3. 'deficiency'; F. 4. 'fabric' F.5. 'economical reason' 2. Subjects perceived design and utility area to be important motives for bed-clothes purchase. 3. Cholor preference of bedclothes was in the order of white pale blue pale green and pale orange. Pattern preference was in the order of stripe plaid solid color floral and polka dot pattern. The combi-nation of patterned fabric and solid color fabric was liked better than the combina-tion of analogic color and the combination of contrasting color. 4. The women in their 20's liked navy blue red stripe plaid pattern and solied color better than 30's and 40's . 40's liked abstract and paisley pattern better then 20's and 30's 5. Low income group lied navy blue and solid color fabric more than the mid-dle and high income group and liked ab-stract pattern less than the middle and high income group. 6. Women in their 20's perceived design to be important motive more than 30's and 40's. 30's perceived brand to be im-portant motive than 20's and 40's 7. High income group perceived design to be important motive more than the middle and low income group. Low in-come group perceived brand and fabric to be less important motives than the middle and high income group, The present findings provide that age and income had a significant effect on bedclothes purchase motives and design preferences of houesewives. The results that white color and the combination of patterned fabric and solid color fabric were liked best indicated that women prefered clean image and chacteristics de-sign of bdeclothes.