• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.43-62
• /
• 2019

At one time, the anomaly detection sector dominated the method of determining whether there was an abnormality based on the statistics derived from specific data. This methodology was possible because the dimension of the data was simple in the past, so the classical statistical method could work effectively. However, as the characteristics of data have changed complexly in the era of big data, it has become more difficult to accurately analyze and predict the data that occurs throughout the industry in the conventional way. Therefore, SVM and Decision Tree based supervised learning algorithms were used. However, there is peculiarity that supervised learning based model can only accurately predict the test data, when the number of classes is equal to the number of normal classes and most of the data generated in the industry has unbalanced data class. Therefore, the predicted results are not always valid when supervised learning model is applied. In order to overcome these drawbacks, many studies now use the unsupervised learning-based model that is not influenced by class distribution, such as autoencoder or generative adversarial networks. In this paper, we propose a method to detect anomalies using generative adversarial networks. AnoGAN, introduced in the study of Thomas et al (2017), is a classification model that performs abnormal detection of medical images. It was composed of a Convolution Neural Net and was used in the field of detection. On the other hand, sequencing data abnormality detection using generative adversarial network is a lack of research papers compared to image data. Of course, in Li et al (2018), a study by Li et al (LSTM), a type of recurrent neural network, has proposed a model to classify the abnormities of numerical sequence data, but it has not been used for categorical sequence data, as well as feature matching method applied by salans et al.(2016). So it suggests that there are a number of studies to be tried on in the ideal classification of sequence data through a generative adversarial Network. In order to learn the sequence data, the structure of the generative adversarial networks is composed of LSTM, and the 2 stacked-LSTM of the generator is composed of 32-dim hidden unit layers and 64-dim hidden unit layers. The LSTM of the discriminator consists of 64-dim hidden unit layer were used. In the process of deriving abnormal scores from existing paper of Anomaly Detection for Sequence data, entropy values of probability of actual data are used in the process of deriving abnormal scores. but in this paper, as mentioned earlier, abnormal scores have been derived by using feature matching techniques. In addition, the process of optimizing latent variables was designed with LSTM to improve model performance. The modified form of generative adversarial model was more accurate in all experiments than the autoencoder in terms of precision and was approximately 7% higher in accuracy. In terms of Robustness, Generative adversarial networks also performed better than autoencoder. Because generative adversarial networks can learn data distribution from real categorical sequence data, Unaffected by a single normal data. But autoencoder is not. Result of Robustness test showed that he accuracy of the autocoder was 92%, the accuracy of the hostile neural network was 96%, and in terms of sensitivity, the autocoder was 40% and the hostile neural network was 51%. In this paper, experiments have also been conducted to show how much performance changes due to differences in the optimization structure of potential variables. As a result, the level of 1% was improved in terms of sensitivity. These results suggest that it presented a new perspective on optimizing latent variable that were relatively insignificant.

• Journal of Intelligence and Information Systems
• /
• v.28 no.1
• /
• pp.89-106
• /
• 2022

Recently, the amount of video data collected from smartphones, CCTVs, black boxes, and high-definition cameras has increased rapidly. According to the increasing video data, the requirements for analysis and utilization are increasing. Due to the lack of skilled manpower to analyze videos in many industries, machine learning and artificial intelligence are actively used to assist manpower. In this situation, the demand for various computer vision technologies such as object detection and tracking, action detection, emotion detection, and Re-ID also increased rapidly. However, the object detection and tracking technology has many difficulties that degrade performance, such as re-appearance after the object's departure from the video recording location, and occlusion. Accordingly, action and emotion detection models based on object detection and tracking models also have difficulties in extracting data for each object. In addition, deep learning architectures consist of various models suffer from performance degradation due to bottlenects and lack of optimization. In this study, we propose an video analysis system consists of YOLOv5 based DeepSORT object tracking model, SlowFast based action recognition model, Torchreid based Re-ID model, and AWS Rekognition which is emotion recognition service. Proposed model uses single-linkage hierarchical clustering based Re-ID and some processing method which maximize hardware throughput. It has higher accuracy than the performance of the re-identification model using simple metrics, near real-time processing performance, and prevents tracking failure due to object departure and re-emergence, occlusion, etc. By continuously linking the action and facial emotion detection results of each object to the same object, it is possible to efficiently analyze videos. The re-identification model extracts a feature vector from the bounding box of object image detected by the object tracking model for each frame, and applies the single-linkage hierarchical clustering from the past frame using the extracted feature vectors to identify the same object that failed to track. Through the above process, it is possible to re-track the same object that has failed to tracking in the case of re-appearance or occlusion after leaving the video location. As a result, action and facial emotion detection results of the newly recognized object due to the tracking fails can be linked to those of the object that appeared in the past. On the other hand, as a way to improve processing performance, we introduce Bounding Box Queue by Object and Feature Queue method that can reduce RAM memory requirements while maximizing GPU memory throughput. Also we introduce the IoF(Intersection over Face) algorithm that allows facial emotion recognized through AWS Rekognition to be linked with object tracking information. The academic significance of this study is that the two-stage re-identification model can have real-time performance even in a high-cost environment that performs action and facial emotion detection according to processing techniques without reducing the accuracy by using simple metrics to achieve real-time performance. The practical implication of this study is that in various industrial fields that require action and facial emotion detection but have many difficulties due to the fails in object tracking can analyze videos effectively through proposed model. Proposed model which has high accuracy of retrace and processing performance can be used in various fields such as intelligent monitoring, observation services and behavioral or psychological analysis services where the integration of tracking information and extracted metadata creates greate industrial and business value. In the future, in order to measure the object tracking performance more precisely, there is a need to conduct an experiment using the MOT Challenge dataset, which is data used by many international conferences. We will investigate the problem that the IoF algorithm cannot solve to develop an additional complementary algorithm. In addition, we plan to conduct additional research to apply this model to various fields' dataset related to intelligent video analysis.

• Korean Journal of Remote Sensing
• /
• v.33 no.6_1
• /
• pp.931-946
• /
• 2017

The purpose of this study is to observe and analyze soil moisture conditions with high resolution and to evaluate its application feasibility to agriculture. For this purpose, we used three Landsat-8 OLI (Operational Land Imager)/TIRS (Thermal Infrared Sensor) optical and thermal infrared satellite images taken from May to June 2015, 2016, and 2017, including the rural areas of Jeollabuk-do, where 46% of agricultural areas are located. The soil moisture conditions at each date in the study area can be effectively obtained through the SPI (Standardized Precipitation Index)3 drought index, and each image has near normal, moderately wet, and moderately dry soil moisture conditions. The temperature vegetation dryness index (TVDI) was calculated to observe the soil moisture status from the Landsat-8 OLI/TIRS images with different soil moisture conditions and to compare and analyze the soil moisture conditions obtained from the SPI3 drought index. TVDI is estimated from the relationship between LST (Land Surface Temperature) and NDVI (Normalized Difference Vegetation Index) calculated from Landsat-8 OLI/TIRS satellite images. The maximum/minimum values of LST according to NDVI are extracted from the distribution of pixels in the feature space of LST-NDVI, and the Dry/Wet edges of LST according to NDVI can be determined by linear regression analysis. The TVDI value is obtained by calculating the ratio of the LST value between the two edges. We classified the relative soil moisture conditions from the TVDI values into five stages: very wet, wet, normal, dry, and very dry and compared to the soil moisture conditions obtained from SPI3. Due to the rice-planing season from May to June, 62% of the whole images were classified as wet and very wet due to paddy field areas which are the largest proportions in the image. Also, the pixels classified as normal were analyzed because of the influence of the field area in the image. The TVDI classification results for the whole image roughly corresponded to the SPI3 soil moisture condition, but they did not correspond to the subdivision results which are very dry, wet, and very wet. In addition, after extracting and classifying agricultural areas of paddy field and field, the paddy field area did not correspond to the SPI3 drought index in the very dry, normal and very wet classification results, and the field area did not correspond to the SPI3 drought index in the normal classification. This is considered to be a problem in Dry/Wet edge estimation due to outlier such as extremely dry bare soil and very wet paddy field area, water, cloud and mountain topography effects (shadow). However, in the agricultural area, especially the field area, in May to June, it was possible to effectively observe the soil moisture conditions as a subdivision. It is expected that the application of this method will be possible by observing the temporal and spatial changes of the soil moisture status in the agricultural area using the optical satellite with high spatial resolution and forecasting the agricultural production.

• Korean Journal of Remote Sensing
• /
• v.36 no.5_2
• /
• pp.867-879
• /
• 2020

As the era of space technology utilization is approaching, the launch of CAS (Compact Advanced Satellite) 500-1/2 satellites is scheduled during 2021 for acquisition of high-resolution images. Accordingly, the increase of image usability and processing efficiency has been emphasized as key design concepts of the CAS 500-1/2 ground station. In this regard, "CAS 500-1/2 Image Acquisition and Utilization Technology Development" project has been carried out to develop core technologies and processing systems for CAS 500-1/2 data collecting, processing, managing and distributing. In this paper, we introduce the results of the above project. We developed an operation system to generate precision images automatically with GCP (Ground Control Point) chip DB (Database) and DEM (Digital Elevation Model) DB over the entire Korean peninsula. We also developed the system to produce ortho-rectified images indexed to 1:5,000 map grids, and hence set a foundation for ARD (Analysis Ready Data)system. In addition, we linked various application software to the operation system and systematically produce mosaic images, DSM (Digital Surface Model)/DTM (Digital Terrain Model), spatial feature thematic map, and change detection thematic map. The major contribution of the developed system and technologies includes that precision images are to be automatically generated using GCP chip DB for the first time in Korea and the various utilization product technologies incorporated into the operation system of a satellite ground station. The developed operation system has been installed on Korea Land Observation Satellite Information Center of the NGII (National Geographic Information Institute). We expect the system to contribute greatly to the center's work and provide a standard for future ground station systems of earth observation satellites.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.12 no.1
• /
• pp.44-48
• /
• 2008

Purpose: In FDG-PET/CT of breast cancer, a sensitivity was 80~96% and a specificity was 75~95% commonly. It was valuable to identify a cancer in early stage been difficult in Mammography. Most of the PET/CT scans have been examined on supine position, so, the image of breast has been acquired by reconstructed whole body scan image. However, using prone position with a compensator, a shape of breast was reassembly shown to be real by gravity. Therefore, the purpose of this study was to evaluate diagnostic value of prone position in FDG PET-CT of breast cancer. Materials and Methods: 30 female patients with doubtful or positive breast cancer were examined. The PET-CT whole body scan was acquired at 60 minutes after $^{18}F$-FDG injection on Supine position. Then, regional breast spot scan was progressed on prone position using a compensator. Each image was evaluated by physicians blinded to patient's data, and statistical analysis did through SUVs measured in PET-CT images. Results: In 27 of 30 patients, prone position was shown accurate discrimination and diagnostic value, but in another 3 patients had a lesion 1cm below, PET-CT couldn't detect it, unlike MRI. Consequently, prone position distinguished a lesion better than Supine position, because of low degree of metamorphosis by gravity. The SUVs analysis of each position was significant (p value=0.004). Conclusion: In PET-CT of breast cancer, prone position could detect micrometastasis as well as primary lesion, better than supine position. Therefore, this study proposes that any technical change considered morphological feature like prone position can offer adequate and useful diagnostic information, together with complementary quantitative analysis.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.20 no.2
• /
• pp.27-31
• /
• 2016

Purpose $^{99m}Tc-DMSA$ renal scan is a test for the comparison of the function by imaging the parenchyma of the kidneys by the cortex of a kidney and by computing the intake ratio of radiation by the left and right kidney. Since the distance between the kidneys and the bladder is not far given the bodily structure of an infant, the bladder is included in the examination domain. Research was carried out with the presumption that counts of bladder would impart an influence on the kidneys at the time of this renal scan. In consideration of the special feature that only a trace amount of a RI is injected in a pediatric examination, research on the method of injection was also carried out concurrently. Materials and Methods With 34 infants aged between 1 month to 12 months for whom a $^{99m}Tc-DMSA$ renal scan was implemented on the subjects, a Post IMAGE was acquired in accordance with the test time after having injected the same quantity of DMSA of 0.5mCi. Then, after having acquired an additional image by shielding the bladder by using a circular lead plate for comparison purposes, a comparison was made by illustrating the percentile of (Lt. Kidney counts + Rt. Kidney counts)/ Total counts, by drawing the same sized ROI (length of 55.2mm X width of 70.0mm). In addition, in the format of a 3-way stopcock, a Heparin cap and direct injection into the patient were performed in accordance with RI injection methods. The differences in the count changes in accordance with each of the methods were compared by injecting an additional 2cc of saline into the 3-way stopcock and Heparin cap. Results The image prior to shielding of the bladder displayed a kidney intake rate with a deviation of $70.9{\pm}3.18%$ while the image after the shielding of the bladder displayed a kidney intake rate with a deviation of $79.4{\pm}5.19%$, thereby showing approximately 6.5~8.5% of difference. In terms of the injection method, the method that used the 3-way form, a deviation of $68.9{\pm}2.80%$ prior to the shielding and a deviation of $78.1{\pm}5.14%$ after the shielding were displayed. In the method of using a Heparin cap, a deviation of $71.3{\pm}5.14%$ prior to the shielding and a deviation of $79.8{\pm}3.26%$ after the shielding were displayed. Lastly, in the method of direct injection into the patient, a deviation of $75.1{\pm}4.30%$ prior to the shielding and a deviation of $82.1{\pm}2.35%$ after the shielding were displayed, thereby illustrating differences in the kidney intake rates in the order of direct injection, a Heparin cap and the 3-way methods. Conclusion Since a substantially minute quantity of radiopharmaceuticals is injected for infants in comparison to adults, the cases of having shielded the bladder by removing radiation of the bladder displayed kidney intake rates that are improved from those of the cases of not having shielded the bladder. Although there are difficulties in securing blood vessels, it is deemed that the method of direct injection would be more helpful in acquisition of better images since it displays improved kidney intake rate in comparison to other methods.

• Journal of Korean Society of Transportation
• /
• v.19 no.4
• /
• pp.35-47
• /
• 2001

Traffic represents one of the largest sources of primary air pollutants in urban area. As a consequence. numerous abatement strategies are being pursued to decrease the ambient concentration of pollutants. A characteristic of most of the these strategies is a requirement for accurate data on both the quantity and spatial distribution of emissions to air in the form of an atmospheric emission inventory database. In the case of traffic pollution, such an inventory must be compiled using activity statistics and emission factors for vehicle types. The majority of inventories are compiled using passive data from either surveys or transportation models and by their very nature tend to be out-of-date by the time they are compiled. The study of current trends are towards integrating urban traffic control systems and assessments of the environmental effects of motor vehicles. In this study, a methodology of motor vehicle emission calculation by using real-time traffic data was studied. A methodology for estimating emissions of CO at a test area in Seoul. Traffic data, which are required on a street-by-street basis, is obtained from induction loops of traffic control system. It was calculated speed-related mass of CO emission from traffic tail pipe of data from traffic system, and parameters are considered, volume, composition, average velocity, link length. And, the result was compared with that of a method of emission calculation by VKT(Vehicle Kilometer Travelled) of vehicles of category.

• Journal of Christian Education in Korea
• /
• v.67
• /
• pp.157-194
• /
• 2021

The issue of AI's ethical consciousness has been constantly on the rise. AI learns and imitates everything behavior human beings do, just like a child. Therefore, the ethical consciousness we currently demand from AI is first the ethical consciousness required of humans, and at the center of it is the dignity of humans. Thus, this study analyzed human identity and its problems according to the development of AI technology, apologized the theological premises and characteristics of human dignity, and sought the direction of human dignity education as follows. First, this study discussed the development of AI and its relation to human beings. The development of AI's technology has led to the sharing of "reason or intelligence" with machines called AI which have been restricted to the exclusive property of mankind. This raised the question of the superior humanity which humans would be remained to be distinguished from AI machines. Second, this study discussed transhumanism and human identity. Transhumanism has been argued for the combination of AI machines and humans in order to improve inefficient human intelligence and human capabilities. However, the combination of AI machines with humans raised the issue of human identity. In the AI era, human identity is to believe thoughts that God had when he built us. Third, this study apologized theological premise and characteristic about human dignity. Human dignity has become a key concept of the constitution and international human rights treaties around the world. Nonetheless, declarative conviction that human is dignified is difficult to be understanded without Christian theological premise. Theological premise of human dignity lies on the fact that human is dignified feature being granted life by Heavenly Father. This feature lies on longing for "Goodness" and "eternality", pursuit of beauty, a happy being in relationship with others. Fourth, this study presented the direction of human dignity education. The direction of human dignity education has to awaken what is identity of human and how human beings were created and how much they are precious. Furthermore, it lead human to ponder consciously and accept the highest value of what human beings are, how they were created, and how precious they are. That is about educating human identity, and its core is that regardless of the circumstances - the wealth gap, knowledge level, skin color, gender, age, disability, etc. - all people are in God's image and for the glory of God, thereby being very important to God.

• Journal of Astronomy and Space Sciences
• /
• v.18 no.1
• /
• pp.33-42
• /
• 2001

In the spectrum of the young bipolar planetary nebula M2-9 obtained from the 1.5m telescope at the Cerro Tololo Inter-American Observatory, we detected the He II feature at 6545 $\AA$ that are proposed to be formed via Raman scattering by atomic hydrogen. However, in the same spectrum, the He II emission lines at 6527 $\AA$ and 6560 $\AA$ are absent, which implies that the He II emission region is hidden from our line of sight and that the H I scattering region is pretty much extended not to be obscured entirely. We performed photoionization computations to estimate the physical size of the He II emission line region to be $10^{16}cm$, from which the location and dimension of the obscuring circumstellar region are inferred and the temperature of the central star must exceed $10^5K$. The angular size of the circumstellar region responsible for the obscuration of the He II emission region is ~1" with the assumption of the distance 01 kpc to M2-9, which is consistent with the recent image of M2-9 obtained with the Hubble Space Telescope.

• Journal of KIISE:Software and Applications
• /
• v.27 no.5
• /
• pp.529-542
• /
• 2000

Recent years, the use of multimedia information is rapidly increasing, and the video media is the most rising one than any others, and this field Integrates all the media into a single data stream. Though the availability of digital video is raised largely, it is very difficult for users to make the effective video access, due to its length and unstructured video format. Thus, the minimal interaction of users and the explicit definition of video structure is a key requirement in the lately developing image and video management systems. This paper defines the terms and hierarchical video structure, and presents the system, which construct the clustering-based video hierarchy, which facilitate users by browsing the summary and do a random access to the video content. Instead of using a single feature and domain-specific thresholds, we use multiple features that have complementary relationship for each other and clustering-based methods that use normalization so as to interact with users minimally. The stage of shot boundary detection extracts multiple features, performs the adaptive filtering process for each features to enhance the performance by eliminating the false factors, and does k-means clustering with two classes. The shot list of a result after the proposed procedure is represented as the video hierarchy by the intelligent unsupervised clustering technique. We experimented the static and the dynamic movie videos that represent characteristics of various video types. In the result of shot boundary detection, we had almost more than 95% good performance, and had also rood result in the video hierarchy.