• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.1
• /
• pp.27-32
• /
• 2019

Research that implements the classification process through Deep Learning algorithm, one of the outstanding human abilities, includes a unimodal model, a multi-modal model, and a multi-modal method using music videos. In this study, the results were better by suggesting a system to analyze each song's spectrum into short samples and vote for the results. Among Deep Learning algorithms, CNN showed superior performance in the category of music genre compared to RNN, and improved performance when CNN and RNN were applied together. The system of voting for each CNN result by Deep Learning a short sample of music showed better results than the previous model and the model with Softmax layer added to the model performed best. The need for the explosive growth of digital media and the automatic classification of music genres in numerous streaming services is increasing. Future research will need to reduce the proportion of undifferentiated songs and develop algorithms for the last category classification of undivided songs.

• Journal of Digital Convergence
• /
• v.19 no.5
• /
• pp.307-313
• /
• 2021

In this paper, a lightweight network with fewer parameters compared to the existing object detection method is proposed. In the case of the currently used detection model, the network complexity has been greatly increased to improve accuracy. Therefore, the proposed network uses EfficientNet as a feature extraction network, and the subsequent layers are formed in a pyramid structure to utilize low-level detailed features and high-level semantic features. An attention process was applied between pyramid structures to suppress unnecessary noise for prediction. All computational processes of the network are replaced by depth-wise and point-wise convolutions to minimize the amount of computation. The proposed network was trained and evaluated using the PASCAL VOC dataset. The features fused through the experiment showed robust properties for various objects through a refinement process. Compared with the CNN-based detection model, detection accuracy is improved with a small amount of computation. It is considered necessary to adjust the anchor ratio according to the size of the object as a future study.

• Journal of Convergence for Information Technology
• /
• v.11 no.7
• /
• pp.31-38
• /
• 2021

In this paper, in order to obtain the optimization of the RNN model used for sentiment analysis, the correlation of each model was studied by observing the trend of loss and accuracy according to hyperparameter tuning. As a research method, after configuring the hidden layer with LSTM and the embedding layer that are most optimized to process sequential data, the loss and accuracy of each model were measured by tuning the unit, batch-size, and embedding size of the LSTM. As a result of the measurement, the loss was 41.9% and the accuracy was 11.4%, and the trend of the optimization model showed a consistently stable graph, confirming that the tuning of the hyperparameter had a profound effect on the model. In addition, it was confirmed that the decision of the embedding size among the three hyperparameters had the greatest influence on the model. In the future, this research will be continued, and research on an algorithm that allows the model to directly find the optimal hyperparameter will continue.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.12
• /
• pp.1835-1845
• /
• 2021

To accurately identify black ice and warn the drivers of information in advance so they can control speed and take preventive measures. In this paper, we propose a lightweight black ice detection network based on infrared road images. A black ice recognition network model based on CNN transfer learning has been developed. Additionally, to further improve the accuracy of black ice recognition, an enhanced lightweight network based on MobileNetV2 has been developed. To reduce the amount of calculation, linear bottlenecks and inverse residuals was used, and four bottleneck groups were used. At the same time, to improve the recognition rate of the model, each bottleneck group was connected to a 3×3 convolutional layer to enhance regional feature extraction and increase the number of feature maps. Finally, a black ice recognition experiment was performed on the constructed infrared road black ice dataset. The network model proposed in this paper had an accurate recognition rate of 99.07% for black ice.

• Journal of the Korea Society for Simulation
• /
• v.30 no.4
• /
• pp.21-29
• /
• 2021

Inverse synthetic aperture radar (ISAR) image is a powerful tool to show the major scattering regions (scatterer points) on the target. It is normally used to identify and classify targets. Finding information for the scatter points of ISAR image plays an important role in modeling the features of targets. In this paper, we propose a modeling method to extract some information about the scatterer points by minimizing approximating error. Here, the extracted information include not only the location of scatterer points but also some statistical data about the error of the their location. These extracted data can be used to implement the randomness of the location of the scatterer points. Furthermore, we reconstruct an image from the extracted data for scatterer points obtained by our proposed method. And we show that the reconstructed ISAR image is well approximated to the original ISAR image in order to justify our proposed modeling method.

• Journal of Space Technology and Applications
• /
• v.1 no.2
• /
• pp.268-281
• /
• 2021

Recently, the synthetic aperture radar (SAR) has been increasingly in demand due to its advantage of being able to observe desired points regardless of time and weather. To utilize SAR data, first of all, many pre-processing such as satellite orbit correction, radiometric calibration, multi-looking, and geocoding are required. For analysis of SAR imagery such as object detection, change detection, and DEM(Digital Elevation Model), additional processings are needed. These pre-processing and additional processes are very complex and require a lot of time and computational resources. In order to handle the SAR images easily, the institutions that use SAR images develop analysis tools and provide users. This paper introduces the function and characteristics of representative SAR imagery analysis tools.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.11
• /
• pp.1486-1494
• /
• 2021

Character recognition is a technology required in various platforms, such as smart parking and text to speech, and many studies are being conducted to improve its performance through new attempts. However, with low-quality image used for character recognition, a difference in resolution of the training image and test image for character recognition occurs, resulting in poor accuracy. To solve this problem, this paper designed an end-to-end learning neural network that combines image super-resolution and character recognition so that the character recognition model performance is robust against various quality data, and implemented an alternative whole learning algorithm to learn the whole neural network. An alternative end-to-end learning and recognition performance test was conducted using the license plate image among various text images, and the effectiveness of the proposed algorithm was verified with the performance test.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.312-318
• /
• 2022

This paper proposes a model which classifies the type of guns and information about sound source location using deep neural network. The proposed classification model is composed of convolutional neural networks (CNN) and long short-term memory (LSTM). For training and test the model, we use the Gunshot Audio Forensic Dataset generated by the project supported by the National Institute of Justice (NIJ). The acoustic signals are transformed to Mel-Spectrogram and they are provided as learning and test data for the proposed model. The model is compared with the control model consisting of convolutional neural networks only. The proposed model shows high accuracy more than 90 %.

• Smart Media Journal
• /
• v.12 no.7
• /
• pp.27-42
• /
• 2023

The development of cloud services and IoT technology has radically changed the cloud environment, and has evolved into a new concept called fog computing and F2C (fog-to-cloud). However, as heterogeneous cloud/fog layers are integrated, problems of access control and security management for end users and edge devices may occur. In this paper, an F2C-based IoT smart health monitoring system architecture was designed to operate a medical information service that can quickly respond to medical emergencies. In addition, a role-based service access control technology was proposed to enhance the security of user's personal health information and sensor information during service interoperability. Through simulation, it was shown that role-based access control is achieved by sharing role registration and user role token issuance information through blockchain. End users can receive services from the device with the fastest response time, and by performing service access control according to roles, direct access to data can be minimized and security for personal information can be enhanced.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.9
• /
• pp.399-406
• /
• 2023

Cloud computing has been evolved to support edge computing architecture that combines fog management layer with edge servers. The main reason why it is received much attention is low communication latency for real-time IoT applications. At the same time, various cloud task scheduling techniques based on artificial intelligence have been proposed. Artificial intelligence-based cloud task scheduling techniques show better performance in comparison to existing methods, but it has relatively high scheduling time. In this paper, we propose a deep learning-based dynamic scheduling with multi-agents supporting scalability in edge computing environments. The proposed method shows low scheduling time than previous artificial intelligence-based scheduling techniques. To show the effectiveness of the proposed method, we compare the performance between previous and proposed methods in a scalable experimental environment. The results show that our method supports real-time IoT applications with low scheduling time, and shows better performance in terms of the number of completed cloud tasks in a scalable experimental environment.