• Journal of Digital Convergence
• /
• v.15 no.2
• /
• pp.173-181
• /
• 2017

In this paper, we have studied tracking as a training stage of considering the position and the scale of a person given its previous position, scale, as well as next and forward image fraction. Unlike other learning methods, CNN is thereby learning combines both time and spatial features from the image for the two consecutive frames. We introduce multiple path ways in CNN to better fuse local and global information. A creative shift-variant CNN architecture is designed so as to alleviate the drift problem when the distracting objects are similar to the target in cluttered environment. Furthermore, we employ CNNs to estimate the scale through the accurate localization of some key points. These techniques are object-independent so that the proposed method can be applied to track other types of object. The capability of the tracker of handling complex situations is demonstrated in many testing sequences. The accuracy of the SVM classifier using the features learnt by the CNN is equivalent to the accuracy of the CNN. This fact confirms the importance of automatically optimized features. However, the computation time for the classification of a person using the convolutional neural network classifier is less than approximately 1/40 of the SVM computation time, regardless of the type of the used features.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.17 no.1
• /
• pp.1-8
• /
• 2016

In this study, an WiFi fingerprinting method based on the k-nn algorithm was applied to improve the accuracy of location tracking of a moving train on a platform and evaluate the performance to minimize the estimation error of location tracking. The data related to the position of the moving train are monitored by the control center for trains and used widely for the safety and comfort of passengers. The train location tracking methods based on WiFi installed by telecom companies were evaluated. In this study, a simulator was developed to consider the environments of two cases; in already installed WiFi devices and new installed WiFi devices. The developed simulator can simulate the localized estimation of the position under a variety of conditions, such as the number of WiFi devices, the area of platform and entry velocity of train. To apply location tracking algorithms, a k-nn algorithm and fuzzy k-nn algorithm were applied selectively according to the underlying condition and also four distance measurement algorithms were applied to compare the error of location tracking. In conclusion, the best method to estimate train location tracking is a combination of the k-nn algorithm and Minkoski distance measurement at a 0.5m grid unit and 8 WiFi AP installed.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.17 no.10
• /
• pp.520-526
• /
• 2016

This study aims to develop a distress prediction model, in order to evaluate the distress prediction power for first-class hotels and to calculate the average financial ratio in the Seoul area by using the financial ratios of hotels in 2015. The sample data was collected from 19 first-class hotels in Seoul and the financial ratios extracted from 14 of these 19 hotels. The results show firstly that the seven financial ratios, viz. the current ratio, total borrowings and bonds payable to total assets, interest coverage ratio to operating income, operating income to sales, net income to stockholders' equity, ratio of cash flows from operating activities to sales and total assets turnover, enable the top-level corporations to be discriminated from the failed corporations and, secondly, by using these seven financial ratios, a discriminant function which classifies the corporations into top-level and failed ones is estimated by linear multiple discriminant analysis. The accuracy of prediction of this discriminant capability turned out to be 87.9%. The accuracy of the estimates obtained by discriminant analysis indicates that the distress prediction model's distress prediction power is 78.95%. According to the analysis results, hotel management groups which administrate low level corporations need to focus on the classification of these seven financial ratios. Furthermore, hotel corporations have very different financial structures and failure prediction indicators from other industries. In accordance with this finding, for the development of credit evaluation systems for such hotel corporations, there is a need for systems to be developed that reflect hotel corporations' financial features.

• Journal of the Korean Chemical Society
• /
• v.63 no.5
• /
• pp.342-345
• /
• 2019

The standards for the contents of glyphosate and glufosinate in foods are specific and well categorized. However, the standard of content in animal feeds is relatively inadequate and the classification is insufficient. There is also constant debate about the risk of glyphosate and glufosinate to human health, but the risk to animals has not been well studied. In this study, we established an analytical method in feeds that is estimated to be the path for animals to ingest glyphosate. The solvent extraction was carried out using 25% methanol. After centrifugation, samples were purified using solid phase extraction (SPE) and quantitatively analysed using LC-MS/MS after concentrated. Assessment of validation was conducted through detection limits, accuracy, and precision tests. The detection limits for the established method were 1.8 of ${\mu}g/kg$ of glufosinate and $2.4{\mu}g/kg$ of glyphosate. Accuracy was ranged from 94.4% to 103.4% and precision was range from 1.5% to 7.2%. Glufosinate was detected in one sample ($ND{\sim}8.8{\mu}g/kg$) and glyphosate was detected in all but one sample ($ND{\sim}337.0{\mu}g/kg$) by applying the analytical method to animal feeds (n=13).

• Convergence Security Journal
• /
• v.21 no.1
• /
• pp.191-200
• /
• 2021

Artificial intelligence technology is developing in an environment where a lot of data is produced due to the development of computing technology, a cloud environment that can store data, and the spread of personal mobile phones. Among these artificial intelligence technologies, the deep neural network provides excellent performance in image recognition and image classification. There have been many studies on image detection for forest fires and fire prevention using such a deep neural network, but studies on detection of cigarette smoking were insufficient. Meanwhile, military units are establishing surveillance systems for various facilities through CCTV, and it is necessary to detect smoking near ammunition stores or non-smoking areas to prevent fires and explosions. In this paper, by reflecting experimentally optimized numerical values such as activation function and learning rate, we did the detection of smoking pictures and non-smoking pictures in two cases. As experimental data, data was constructed by crawling using pictures of smoking and non-smoking published on the Internet, and a machine learning library was used. As a result of the experiment, when the learning rate is 0.004 and the optimization algorithm Adam is used, it can be seen that the accuracy of 93% and F1-score of 94% are obtained.

• Journal of Platform Technology
• /
• v.9 no.3
• /
• pp.52-62
• /
• 2021

Quality control is critical at manufacturing sites and is key to predicting the risk of quality defect before manufacturing. However, the reliability of manual quality control methods is affected by human and physical limitations because manufacturing processes vary across industries. These limitations become particularly obvious in domain areas with numerous manufacturing processes, such as the manufacture of major nuclear equipment. This study proposed a novel method for predicting the risk of quality defects by using natural language processing and machine learning. In this study, production data collected over 6 years at a factory that manufactures main equipment that is installed in nuclear power plants were used. In the preprocessing stage of text data, a mapping method was applied to the word dictionary so that domain knowledge could be appropriately reflected, and a hybrid algorithm, which combined n-gram, Term Frequency-Inverse Document Frequency, and Singular Value Decomposition, was constructed for sentence vectorization. Next, in the experiment to classify the risky processes resulting in poor quality, k-fold cross-validation was applied to categorize cases from Unigram to cumulative Trigram. Furthermore, for achieving objective experimental results, Naive Bayes and Support Vector Machine were used as classification algorithms and the maximum accuracy and F1-score of 0.7685 and 0.8641, respectively, were achieved. Thus, the proposed method is effective. The performance of the proposed method were compared and with votes of field engineers, and the results revealed that the proposed method outperformed field engineers. Thus, the method can be implemented for quality control at manufacturing sites.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.24 no.2
• /
• pp.52-63
• /
• 2021

In previous research, the coastal wetlands were detected by using the vegetation indices or land cover classification maps derived from the multispectral bands of the satellite or aerial imagery, and this approach caused the various limitations for detecting the coastal wetlands with high accuracy due to the difficulty of acquiring both land cover and topographic information by using the single remote sensing data. This research suggested the efficient methodology for detecting the coastal wetlands using the sentinel-2 satellite image and SRTM(Shuttle Radar Topography Mission) DEM (Digital Elevation Model) acquired in Gomsoman Bay, west coasts of South Korea through the following steps. First, the NDWI(Normalized Difference Water Index) image was generated using the green and near-infrared bands of the given Sentinel-2 satellite image. Then, the binary image that separating lands and waters was generated from the NDWI image based on the pixel intensity value 0.2 as the threshold and the other binary image that separating the upper sea level areas and the under sea level areas was generated from the SRTM DEM based on the pixel intensity value 0 as the threshold. Finally, the coastal wetland map was generated by overlaying analysis of these binary images. The generated coastal wetland map had the 94% overall accuracy. In addition, the other types of wetlands such as inland wetlands or mountain wetlands were not detected in the generated coastal wetland map, which means that the generated coastal wetland map can be used for the coastal wetland management tasks.

• Journal of Internet Computing and Services
• /
• v.21 no.6
• /
• pp.13-21
• /
• 2020

The Android author identification study can be interpreted as a method for revealing the source in a narrow range, but if viewed in a wide range, it can be interpreted as a study to gain insight to identify similar works through known works. The problem found in the Android author identification study is that it is an important code on the Android system, but it is difficult to find the important feature of the author due to the meaningless codes. Due to this, legitimate codes or behaviors were also incorrectly defined as malicious codes. To solve this, we introduced the concept of survival network to solve the problem by removing the features found in various Android apps and surviving unique features defined by authors. We conducted an experiment comparing the proposed framework with a previous study. From the results of experiments on 440 authors' identified apps, we obtained a classification accuracy of up to 92.10%, and showed a difference of up to 3.47% from the previous study. It used a small amount of learning data, but because it used unique features without duplicate features for each author, it was considered that there was a difference from previous studies. In addition, even in comparative experiments with previous studies according to the feature definition method, the same accuracy can be shown with a small number of features, and this can be seen that continuously overlapping meaningless features can be managed through the concept of a survival network.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.3
• /
• pp.133-139
• /
• 2021

Currently, the process of constructing a high-definition road map has a high proportion of manual labor, so there are limitations in construction time and cost. Research to automate map production with high-definition road maps using artificial intelligence is being actively conducted, but since the construction of training data for the map construction is also done manually, there is a need to automatically build training data. Therefore, in this study, after converting to images using point clouds acquired by a mobile mapping system, the road marking areas were extracted through image reclassification and overlap analysis using thresholds. Then, a methodology was proposed to automatically construct training data for deep learning data for the high-definition road map through the classification of the polygon types in the extracted regions. As a result of training 2,764 lane data constructed through the proposed methodology on a deep learning-based PointNet model, the training accuracy was 99.977%, and as a result of predicting the lanes of three color types using the trained model, the accuracy was 99.566%. Therefore, it was found that the methodology proposed in this study can efficiently produce training data for high-definition road maps, and it is believed that the map production process of road markings can also be automated.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.2
• /
• pp.127-133
• /
• 2021

A recent study of people physically falling focused on analyzing the motions of the falls using a recurrent neural network (RNN) and a deep learning approach to get good results from detecting 2D human poses from a single color image. In this paper, we investigate a detection method for estimating the position of the head and shoulder keypoints and the acceleration of positional change using the skeletal keypoints information extracted using PoseNet from an image obtained with a low-cost 2D RGB camera, increasing the accuracy of judgments about the falls. In particular, we propose a fall detection method based on the characteristics of post-fall posture in the fall motion-analysis method. A public data set was used to extract human skeletal features, and as a result of an experiment to find a feature extraction method that can achieve high classification accuracy, the proposed method showed a 99.8% success rate in detecting falls more effectively than a conventional, primitive skeletal data-use method.