• Journal of Broadcast Engineering
• /
• v.28 no.3
• /
• pp.267-274
• /
• 2023

There is a problem that the size of the dataset is insufficient due to the limitation of the collection of the medical image public data, so there is a possibility that the existing studies are overfitted to the public dataset. In this paper, we compare the performance of eight (Unet, X-Net, HarDNet, SegNet, PSPNet, SwinUnet, 3D-ResU-Net, UNETR) medical image semantic segmentation models to revalidate the superiority of existing models. Anatomical Tracings of Lesions After Stroke (ATLAS) V1.2, a public dataset for stroke diagnosis, is used to compare the performance of the models and the performance of the models in ATLAS V2.0. Experimental results show that most models have similar performance in V1.2 and V2.0, but X-net and 3D-ResU-Net have higher performance in V1.2 datasets. These results can be interpreted that the models may be overfitted to V1.2.

• Journal of Korean Society of Archives and Records Management
• /
• v.18 no.2
• /
• pp.109-133
• /
• 2018

The need for the records management of administrative information dataset has led to a broad consensus among archivists and has been continuously studied. In the meantime, information technology has greatly advanced, and the development and redevelopment of information management systems have been increasing. Nevertheless, dataset management in information management system has not been practiced in public organizations. This is because it is supposed that no practical management plan exists. From the point of view that practical dataset management methods should be based on the reality of dataset creation and management environment, this study investigates various active datasets in working administrative information systems. The examples and the information drawn from the examination are expected to contribute to dataset management planning. Moreover, the research methods can be utilized in further studies.

• Journal of Korea Multimedia Society
• /
• v.22 no.10
• /
• pp.1133-1141
• /
• 2019

Most of the previous studies for bark recognition have focused on the extraction of LBP-like statistical features. Deep learning approach was not well studied because of the difficulty of acquiring large volume of bark image dataset. To overcome the bark dataset problem, this study utilizes the MobileNet which was trained with the ImageNet dataset. This study proposes two approaches. One is to extract features by the pixel-wise convolution and classify the features with SVM. The other is to tune the weights of the MobileNet by flexibly freezing layers. The experimental results with two public bark datasets, BarkTex and Trunk12, show that the proposed methods are effective in bark recognition. Especially the results of the flexible tunning method outperform state-of-the-art methods. In addition, it can be applied to mobile devices because the MobileNet is compact compared to other deep learning models.

• The Korean Journal of Archival Studies
• /
• no.19
• /
• pp.251-291
• /
• 2009

Due to the trend toward computerization of business services in public sector and the push for e-government, the volume of records that are produced in electronic system and the types of records vary as well. Of those types, dataset is attracting everyone's attention because it is rapidly being supplied. Even though the administrative information system stipulated as an electronic record production system is increasing in number, as it is in blind spot for records management, the system can be superannuated or the records can be lost in case new system is developed. In addition, the system was designed not considering records management, it is managed in an unsatisfactory state because of not meeting the features and quality requirements as records management system. In the advanced countries, they recognized the importance of dataset and then managed the archives for dataset and carried out the project on management systems and a preservation formats for keeping data. Korea also is carrying out the researches on an dataset and individual administrative information systems, but the official scheme has not been established yet. In this study the items for managing archives which should be reflected when the administrative information system is designed was offered in two respects - an identification method and a quality requirement. The major directions for this system are as follows. First, as the dataset is a kind of an electronic record, it is necessary to reflect this factor from the design step prior to production. Second, the system should be established integrating the strategy for records management to the information strategy for the whole organization. In this study, based on such two directions the strategies to establish the identification for dataset in a frame to push e-government were suggested. The problem on the archiving steps including preservation format and the management procedures in dataset archive does not included in the research contents. In line with this, more researches on those contents as well as a variety of researches on dataset are expected to be more actively conducted.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.209-211
• /
• 2022

YouTube, a video platform used by millions of people worldwide, provides a rapidly growing video service. This study aims to understand the characteristics and cultural differences of each country using the Kaggle dataset, one of the public datasets, and to show the usefulness of the public dataset. For this purpose, we analyze data from 11 countries, 15 categories, and about 1.1 million trending videos. This study adopts Python to obtain the number of videos by category for data analysis, the selection period of videos rapidly increasing in popularity, and the ratio of unique videos. In the future, based on machine learning, we plan to research to help diagnose individual videos and establish channel operation plans and strategies by predicting the selection possibility and selection period based on machine learning.

• Journal of Information Technology Services
• /
• v.15 no.3
• /
• pp.223-245
• /
• 2016

This study defines copyrighted public records in broad sense including open government data and public domain except for some private records. Additionally, this study aims to investigate improvement plan for maximizing utilization of copyrighted public records in web-sites using customer side, without consideration of supplier side. For this purpose, qualitative study method was used with grounded theory on analyzed problems from literature review and case study. Literature review was concentrated on definition of open data and abroad utilization indicators whereas case study analyzed current situation of four web-sites providing copyrighted public records. Converged opinions from in-depth interview and various statistical data was analyzed as a basis for grounded theory, then a paradigm model was constructed and future improvement plans were presented. The findings imply that opening of copyrighted public records is not just important for quantitative results, rather it requires qualitative improvement providing latest credible information that is consistent with the demand of the customer. Thus, development of service platform and business models for copyrighted public records are urgent task.

• The Journal of Asian Finance, Economics and Business
• /
• v.9 no.5
• /
• pp.345-354
• /
• 2022

The importance of the public sector's role in fostering a positive business climate has prompted public sector organizations to consistently enhance their performance. The study aims to develop service orientation dimensions for public sectors and examine the relationship between service orientation and organizational performance. A field survey was employed in this study. Six hundred questionnaires were distributed, and four hundred and eighty-eight were returned and analyzed. Factor analysis and multiple regression analysis were used in the dataset. This study identifies five dimensions of organizational service orientation in public sector service organizations: technology-service standard-communication, service vision, service delivery, service training and powering, and servant leadership. The result also concludes that service orientation influences organizational performance, such as corporate growth, service quality image, IT effectiveness, service innovation, and public complaint. This study's findings imply that public sector organizations should rectify service orientation factors to increase corporate growth, service quality image, IT effectiveness, service innovation, and public complaint reduction. Managerial guidelines are presented for developing a service orientation.

• Journal of Korea Multimedia Society
• /
• v.24 no.7
• /
• pp.880-889
• /
• 2021

In everyday life, recognizing people's emotions from their frames is essential and is a popular research domain in the area of computer vision. Visual emotion has a severe class imbalance in which most of the data are distributed in specific categories. The existing methods do not consider class imbalance and used accuracy as the performance metric, which is not suitable for evaluating the performance of the imbalanced dataset. Therefore, we proposed a method for recognizing visual emotion using balanced data augmentation to address the class imbalance. The proposed method generates a balanced dataset by adopting the random over-sampling and image transformation methods. Also, the proposed method uses the Focal loss as a loss function, which can mitigate the class imbalance by down weighting the well-classified samples. EfficientNet, which is the state-of-the-art method for image classification is used to recognize visual emotion. We compare the performance of the proposed method with that of conventional methods by using a public dataset. The experimental results show that the proposed method increases the F1 score by 40% compared with the method without data augmentation, mitigating class imbalance without loss of classification accuracy.

• Korean Journal of Remote Sensing
• /
• v.37 no.4
• /
• pp.747-761
• /
• 2021

With the increasing application of UAVs in intelligent transportation systems, vehicle detection for aerial images has become an essential engineering technology and has academic research significance. In this paper, a vehicle detection method for aerial images based on the YOLOv4 deep learning algorithm is presented. At present, the most known datasets are VOC (The PASCAL Visual Object Classes Challenge), ImageNet, and COCO (Microsoft Common Objects in Context), which comply with the vehicle detection from UAV. An integrated dataset not only reflects its quantity and photo quality but also its diversity which affects the detection accuracy. The method integrates three public aerial image datasets VAID, UAVD, DOTA suitable for YOLOv4. The training model presents good test results especially for small objects, rotating objects, as well as compact and dense objects, and meets the real-time detection requirements. For future work, we will integrate one more aerial image dataset acquired by our lab to increase the number and diversity of training samples, at the same time, while meeting the real-time requirements.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.2
• /
• pp.73-81
• /
• 2024

Recently, mobile manipulators have been utilized in agriculture industry for weed removal and harvest automation. This paper proposes a domain adaptive fruit detection method for harvest automation, by utilizing OWL-ViT model which is an open-vocabulary object detection model. The vision-language model can detect objects based on text prompt, and therefore, it can be extended to detect objects of undefined categories. In the development of deep learning models for real-world problems, constructing a large-scale labeled dataset is a time-consuming task and heavily relies on human effort. To reduce the labor-intensive workload, we utilized a large-scale public dataset as a source domain data and employed a domain adaptation method. Adversarial learning was conducted between a domain discriminator and feature extractor to reduce the gap between the distribution of feature vectors from the source domain and our target domain data. We collected a target domain dataset in a real-like environment and conducted experiments to demonstrate the effectiveness of the proposed method. In experiments, the domain adaptation method improved the AP50 metric from 38.88% to 78.59% for detecting objects within the range of 2m, and we achieved 81.7% of manipulation success rate.