• Journal of Animal Science and Technology
• /
• v.63 no.2
• /
• pp.367-379
• /
• 2021

The objectives of this study were to evaluate convolutional neural network models and computer vision techniques for the classification of swine posture with high accuracy and to use the derived result in the investigation of the effect of dietary fiber level on the behavioral characteristics of the pregnant sow under low and high ambient temperatures during the last stage of gestation. A total of 27 crossbred sows (Yorkshire × Landrace; average body weight, 192.2 ± 4.8 kg) were assigned to three treatments in a randomized complete block design during the last stage of gestation (days 90 to 114). The sows in group 1 were fed a 3% fiber diet under neutral ambient temperature; the sows in group 2 were fed a diet with 3% fiber under high ambient temperature (HT); the sows in group 3 were fed a 6% fiber diet under HT. Eight popular deep learning-based feature extraction frameworks (DenseNet121, DenseNet201, InceptionResNetV2, InceptionV3, MobileNet, VGG16, VGG19, and Xception) used for automatic swine posture classification were selected and compared using the swine posture image dataset that was constructed under real swine farm conditions. The neural network models showed excellent performance on previously unseen data (ability to generalize). The DenseNet121 feature extractor achieved the best performance with 99.83% accuracy, and both DenseNet201 and MobileNet showed an accuracy of 99.77% for the classification of the image dataset. The behavior of sows classified by the DenseNet121 feature extractor showed that the HT in our study reduced (p < 0.05) the standing behavior of sows and also has a tendency to increase (p = 0.082) lying behavior. High dietary fiber treatment tended to increase (p = 0.064) lying and decrease (p < 0.05) the standing behavior of sows, but there was no change in sitting under HT conditions.

• Journal of Intelligence and Information Systems
• /
• v.29 no.3
• /
• pp.419-437
• /
• 2023

As the number of thin filers in Korea surpasses 12 million, there is a growing interest in enhancing the accuracy of assessing their credit default risk to generate additional revenue. Specifically, researchers are actively pursuing the development of default prediction models using machine learning and deep learning algorithms, in contrast to traditional statistical default prediction methods, which struggle to capture nonlinearity. Among these efforts, Graph Neural Network (GNN) architecture is noteworthy for predicting default in situations with limited data on thin filers. This is due to their ability to incorporate network information between borrowers alongside conventional credit-related data. However, prior research employing graph neural networks has faced limitations in effectively handling diverse categorical variables present in credit information. In this study, we introduce the Transformer embedded Graph Convolutional Network (TeGCN), which aims to address these limitations and enable effective default prediction for thin filers. TeGCN combines the TabTransformer, capable of extracting contextual information from categorical variables, with the Graph Convolutional Network, which captures network information between borrowers. Our TeGCN model surpasses the baseline model's performance across both the general borrower dataset and the thin filer dataset. Specially, our model performs outstanding results in thin filer default prediction. This study achieves high default prediction accuracy by a model structure tailored to characteristics of credit information containing numerous categorical variables, especially in the context of thin filers with limited data. Our study can contribute to resolving the financial exclusion issues faced by thin filers and facilitate additional revenue within the financial industry.

• Journal of the Korean Society for Library and Information Science
• /
• v.57 no.2
• /
• pp.435-452
• /
• 2023

This study experimented with automatic classification of subject headings using BERT-based transfer learning model, and analyzed its performance. This study analyzed the classification performance according to the main class of KDC classification and the category type of subject headings. Six datasets were constructed from Korean national bibliographies based on the frequency of the assignments of subject headings, and titles were used as classification features. As a result, classification performance showed values of 0.6059 and 0.5626 on the micro F1 and macro F1 score, respectively, in the dataset (1,539,076 records) containing 3,506 subject headings. In addition, classification performance by the main class of KDC classification showed good performance in the class General works, Natural science, Technology and Language, and low performance in Religion and Arts. As for the performance by the category type of the subject headings, the categories of plant, legal name and product name showed high performance, whereas national treasure/treasure category showed low performance. In a large dataset, the ratio of subject headings that cannot be assigned increases, resulting in a decrease in final performance, and improvement is needed to increase classification performance for low-frequency subject headings.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.3
• /
• pp.183-191
• /
• 2021

Non-intrusive load monitoring is a technology that can be used for predicting and classifying the type of appliances through real-time monitoring of user power consumption, and it has recently got interested as a means of energy-saving. In this paper, we propose a system for classifying appliances from user consumption data by combining GAF(Gramian angular field) technique that can be used for converting one-dimensional data to the two-dimensional matrix with convolutional neural networks. We use REDD(residential energy disaggregation dataset) that is the public appliances power data and confirm the classification accuracy of the GASF(Gramian angular summation field) and GADF(Gramian angular difference field). Simulation results show that both models showed 94% accuracy on appliances with binary-state(on/off) and that GASF showed 93.5% accuracy that is 3% higher than GADF on appliances with multi-state. In later studies, we plan to increase the dataset and optimize the model to improve accuracy and speed.

• KIISE Transactions on Computing Practices
• /
• v.23 no.3
• /
• pp.183-187
• /
• 2017

Smart phone users prefer fast reading and texting. Hence, users frequently use abbreviated sequences of words and phrases. Nowadays, abbreviations are widely used from chat terms to technical terms. Therefore, gathering abbreviations would be helpful to many services, including information retrieval, recommendation system, and so on. However, manually gathering abbreviations needs to much effort and cost. This is because new abbreviations are continuously generated whenever a new material such as a TV program or a phenomenon is made. Thus it is required to generate of abbreviations automatically. To generate Korean abbreviations, the existing methods use the rule-based approach. The rule-based approach has limitations, in that it is unable to generate irregular abbreviations. Another problem is to decide the correct abbreviation among candidate abbreviations generated rules. To address the limitations, we propose a method of generating Korean abbreviations automatically using sequence-to-sequence learning in this paper. The sequence-to-sequence learning can generate irregular abbreviation and does not lead to the problem of deciding correct abbreviation among candidate abbreviations. Accordingly, it is suitable for generating Korean abbreviations. To evaluate the proposed method, we use dataset of two type. As experimental results, we prove that our method is effective for irregular abbreviations.

• The Journal of the Korea Contents Association
• /
• v.21 no.11
• /
• pp.135-144
• /
• 2021

Recently, data-driven decision-making technology has become a key technology leading the data industry, and machine learning technology for this requires high-quality training datasets. However, real-world data contains missing values for various reasons, which degrades the performance of prediction models learned from the poor training data. Therefore, in order to build a high-performance model from real-world datasets, many studies on automatically imputing missing values in initial training data have been actively conducted. Many of conventional machine learning-based imputation techniques for handling missing data involve very time-consuming and cumbersome work because they are applied only to numeric type of columns or create individual predictive models for each columns. Therefore, this paper proposes a new data imputation technique called 'Denoising Self-Attention Network (DSAN)', which can be applied to mixed-type dataset containing both numerical and categorical columns. DSAN can learn robust feature expression vectors by combining self-attention and denoising techniques, and can automatically interpolate multiple missing variables in parallel through multi-task learning. To verify the validity of the proposed technique, data imputation experiments has been performed after arbitrarily generating missing values for several mixed-type training data. Then we show the validity of the proposed technique by comparing the performance of the binary classification models trained on imputed data together with the errors between the original and imputed values.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.3
• /
• pp.166-172
• /
• 2022

In this study, we photograph driving obstacle objects such as crosswalks, side spheres, manholes, braille blocks, partial ramps, temporary safety barriers, stairs, and inclined curb that hinder or cause inconvenience to the movement of the vulnerable using electric mobility aids. We develop an optimal AI model that classifies photographed objects and automatically recognizes them, and implement an algorithm that can efficiently determine obstacles in front of electric mobility aids. In order to enable object detection to be AI learning with high probability, the labeling form is labeled as a polygon form when building a dataset. It was developed using a Mask R-CNN model in Detectron2 framework that can detect objects labeled in the form of polygons. Image acquisition was conducted by dividing it into two groups: the general public and the transportation weak, and image information obtained in two areas of the test bed was secured. As for the parameter setting of the Mask R-CNN learning result, it was confirmed that the model learned with IMAGES_PER_BATCH: 2, BASE_LEARNING_RATE 0.001, MAX_ITERATION: 10,000 showed the highest performance at 68.532, so that the user can quickly and accurately recognize driving risks and obstacles.

• Journal of the Institute of Convergence Signal Processing
• /
• v.24 no.2
• /
• pp.119-125
• /
• 2023

In this study, we photograph driving obstacle objects such as crosswalks, side spheres, manholes, braille blocks, partial ramps, temporary safety barriers, stairs, and inclined curb that hinder or cause inconvenience to the movement of the vulnerable using electric mobility aids. We develop an optimal AI model that classifies photographed objects and automatically recognizes them, and implement an algorithm that can efficiently determine obstacles in front of electric mobility aids. In order to enable object detection to be AI learning with high probability, the labeling form is labeled as a polygon form when building a dataset. It was developed using a Mask R-CNN model in Detectron2 framework that can detect objects labeled in the form of polygons. Image acquisition was conducted by dividing it into two groups: the general public and the transportation weak, and image information obtained in two areas of the test bed was secured. As for the parameter setting of the Mask R-CNN learning result, it was confirmed that the model learned with IMAGES_PER_BATCH: 2, BASE_LEARNING_RATE 0.001, MAX_ITERATION: 10,000 showed the highest performance at 68.532, so that the user can quickly and accurately recognize driving risks and obstacles.

• 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.

• Journal of Internet Computing and Services
• /
• v.20 no.6
• /
• pp.55-63
• /
• 2019

Recently, the use of real-world image data has been increasing to express realistic virtual environments in various application fields such as education, manufacturing, and construction. In particular, with increasing interest in digital twins like smart cities, realistic 3D urban models are being built using real-world images, such as aerial images. However, the captured aerial image includes shadows from the sun, and the 3D city model including the shadows has a problem of distorting and expressing information to the user. Many studies have been conducted to remove the shadow, but it is recognized as a challenging problem that is still difficult to solve. In this paper, we construct a virtual environment dataset including the height map of buildings using 3D spatial information provided by VWorld, and We propose a new shadow detection method using height map and deep learning. According to the experimental results, We can observed that the shadow detection error rate is reduced when using the height map.