• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.46 no.1
• /
• pp.55-67
• /
• 2023

Lightweight face recognition models, as one of the most popular and long-standing topics in the field of computer vision, has achieved vigorous development and has been widely used in many real-world applications due to fewer number of parameters, lower floating-point operations, and smaller model size. However, few surveys reviewed lightweight models and reimplemented these lightweight models by using the same calculating resource and training dataset. In this survey article, we present a comprehensive review about the recent research advances on the end-to-end efficient lightweight face recognition models and reimplement several of the most popular models. To start with, we introduce the overview of face recognition with lightweight models. Then, based on the construction of models, we categorize the lightweight models into: (1) artificially designing lightweight FR models, (2) pruned models to face recognition, (3) efficient automatic neural network architecture design based on neural architecture searching, (4) Knowledge distillation and (5) low-rank decomposition. As an example, we also introduce the SqueezeFaceNet and EfficientFaceNet by pruning SqueezeNet and EfficientNet. Additionally, we reimplement and present a detailed performance comparison of different lightweight models on the nine different test benchmarks. At last, the challenges and future works are provided. There are three main contributions in our survey: firstly, the categorized lightweight models can be conveniently identified so that we can explore new lightweight models for face recognition; secondly, the comprehensive performance comparisons are carried out so that ones can choose models when a state-of-the-art end-to-end face recognition system is deployed on mobile devices; thirdly, the challenges and future trends are stated to inspire our future works.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.2-5
• /
• 2022

Human organs in the body have a complex structure, and in particular, the small intestine is about 7m long, so endoscopy is not easy and the risk of endoscopy is high. Currently, the test is performed with a capsule endoscope, and the test time is very long. The doctor connects the removed storage device to the computer to store the patient's capsule endoscope image and reads it using a program, but the capsule endoscope test results in a long image length, which takes a lot of time to read. In addition, in the case of the small intestine, there are many curves due to villi, so the occlusion area or light and shade of the image are clearly visible during the examination, and there may be cases where lesions and abnormal signs are missed during the examination. In this paper, we provide a method of assisting small intestine capsule endoscopic lesion examination using artificial neural networks to shorten the doctor's image reading time and improve diagnostic reliability.

• Journal of architectural history
• /
• v.32 no.6
• /
• pp.7-21
• /
• 2023

This research aims to understand the specific aspects of the utilization of the pavilion by a group of governors in the mobile office system of the early Joseon Dynasty through two diaries written in the 16th century. Miam Diary by Yu Hee-chun, a governor of Jeolla Province, and Jaeyeongnam Diary by Hwang Sa-woo, a chief aide of Gyeongsang Province, are important historical materials that reveal the utilization patterns of the pavilion by the governor, who was the decision maker and main user of governmental pavilions. As a result of analyzing the two diaries, the utilization of governmental pavilions was concentrated in the hot summer season, May to July, which is closely related to the perception of temperature and humidity. While pavilions are mostly used as office and banquet places, some notable usage patterns have been identified. When there were several governmental pavilions in a town, the order of appreciation was determined by considering the location and scenery, and the pavilions were also used as a place to encourage learning as governors taught Confucian scholars well. Governmental pavilions functioned as a device to visualize hierarchy through seating and accommodation arrangements. The authors of the diaries left comments on the famous pavilions and sometimes went to see the pavilions after asking for permission from the superior. This research is meaningful in that it reconstructed the relationship network and phases of the times of governmental pavilions scattered across the country through institutions and daily life.

• Smart Structures and Systems
• /
• v.33 no.1
• /
• pp.55-64
• /
• 2024

This paper proposes a fusion imaging-based coating-defect classification method for steel structures that uses zero-shot learning. In the proposed method, a halogen lamp generates heat energy on the coating surface of a steel structure, and the resulting heat responses are measured by an infrared (IR) camera, while photos of the coating surface are captured by a charge-coupled device (CCD) camera. The measured heat responses and visual images are then analyzed using zero-shot learning to classify the coating defects, and the estimated coating defects are visualized throughout the inspection surface of the steel structure. In contrast to older approaches to coating-defect classification that relied on visual inspection and were limited to surface defects, and older artificial neural network (ANN)-based methods that required large amounts of data for training and validation, the proposed method accurately classifies both internal and external defects and can classify coating defects for unobserved classes that are not included in the training. Additionally, the proposed model easily learns about additional classifying conditions, making it simple to add classes for problems of interest and field application. Based on the results of validation via field testing, the defect-type classification performance is improved 22.7% of accuracy by fusing visual and thermal imaging compared to using only a visual dataset. Furthermore, the classification accuracy of the proposed method on a test dataset with only trained classes is validated to be 100%. With word-embedding vectors for the labels of untrained classes, the classification accuracy of the proposed method is 86.4%.

• Journal of Information Technology Services
• /
• v.23 no.1
• /
• pp.1-10
• /
• 2024

The IT environment is changing due to the acceleration of digital transformation in enterprises and organizations. This expansion of the digital space makes centralized cybersecurity controls more difficult. For this reason, cyberattacks are increasing in frequency and severity and are becoming more sophisticated, such as ransomware and digital supply chain attacks. Even in large organizations with numerous security personnel and systems, security incidents continue to occur due to unmanaged and unknown threats and vulnerabilities to IT assets. It's time to move beyond the current focus on detecting and responding to security threats to managing the full range of cyber risks. This requires the implementation of asset Inventory for comprehensive management by collecting and integrating all IT assets of the enterprise and organization in a wide range. IT Asset Management(ITAM) systems exist to identify and manage various assets from a financial and administrative perspective. However, the asset information managed in this way is not complete, and there are problems with duplication of data. Also, it is insufficient to update of data-set, including Network Infrastructure, Active Directory, Virtualization Management, and Cloud Platforms. In this study, we, the researcher group propose a new framework for automated 'Comprehensive IT Asset Management(CITAM)' required for security operations by designing a process to automatically collect asset data-set. Such as the Hostname, IP, MAC address, Serial, OS, installed software information, last seen time, those are already distributed and stored in operating IT security systems. CITAM framwork could classify them into unique device units through analysis processes in term of aggregation, normalization, deduplication, validation, and integration.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.1
• /
• pp.189-196
• /
• 2024

Frequent false positives alarm from the Intelligent Selective Control System have raised significant concerns. These persistent issues have led to declines in operational efficiency and market credibility among agents. Developing a new model or replacing the existing one to mitigate false positives alarm entails substantial opportunity costs; hence, improving the quality of the training dataset is pragmatic. However, smaller organizations face challenges with inadequate capabilities in dataset collection and refinement. This paper proposes an automatic human pose data collection system centered around a human pose estimation model, utilizing camera-based sensor fusion techniques and edge devices. The system facilitates the direct collection and real-time processing of field data at the network periphery, distributing the computational load that typically centralizes. Additionally, by directly labeling field data, it aids in constructing new training datasets.

• Smart Media Journal
• /
• v.13 no.4
• /
• pp.76-85
• /
• 2024

In this paper, we propose a system that improves the data integrity of IoT(Internet of Things) devices in the virtual environment by combining Hyperledger Indy and MQTT(Message Queuing Telemetry Transport). The system complements the limitations of the centralized system by realizing a DPKI(Decentralized Public Key Infrastructure) structure that utilizes a distributed network in publish-subscribe(pub/sub) pattern communication. Digital signature technology was applied to ensure the data integrity of IoT devices and communication scenarios between the four core components of the client, IoT device, broker, and blockchain, as well as a topic structure using a decentralized identifier to ensure safety in the virtual environment. We present a systematic method for transparent data exchange. To prove the performance of the proposed system, this paper conducted experiments on four scenarios and evaluated communication performance in a virtual environment. The experimental results confirmed that the proposed system provides a reliable IoT data communication structure in a virtual environment.

• Journal of the Korea Computer Graphics Society
• /
• v.30 no.3
• /
• pp.87-97
• /
• 2024

This study aims to develop a system that enables users employing PC and VR devices to collaboratively analyze data visualizations in a remote environment. The system provides a task-oriented node-link control interface to aid users in understanding the visualization analysis process and effectively distributing roles. Additionally, it offers a network environment where multiple users can collaborate and receive feedback on visualization analysis even when physically separated. To elucidate the collaborative analysis method implemented in the system, we designed a scenario. Furthermore, we conducted a pilot experiment to evaluate the system's usability with participants majoring in related fields. The experimental results confirmed that users can freely analyze data through easily comprehensible interface manipulations in an extended reality space, and efficiently conduct real-time collaborative analysis in a remote environment.

• Journal of IKEEE
• /
• v.18 no.4
• /
• pp.552-558
• /
• 2014

This paper suggests how to design a ZigBee-chip-based communication module to remotely measure radiation level. The suggested communication module consists of two control processors for the chip as generally required to configure a ZigBee system, and one chip module to configure a ZigBee RF device. The ZigBee-chip-based communication module for remote radiation measurement consists of a wireless communication controller; sensor and high-voltage generator; charger and power supply circuit; wired communication part; and RF circuit and antenna. The wireless communication controller is to control wireless communication for ZigBee and to measure radiation level remotely. The sensor and high-voltage generator generates 500 V in two consecutive series to amplify and filter pulses of radiation detected by G-M Tube. The charger and power supply circuit part is to charge lithium-ion battery and supply power to one-chip processors. The wired communication part serves as a RS-485/422 interface to enable USB interface and wired remote communication for interfacing with PC and debugging. RF circuit and antenna applies an RLC passive component for chip antenna to configure BALUN and antenna impedance matching circuit, allowing wireless communication. After configuring the ZigBee-chip-based communication module, tests were conducted to measure radiation level remotely: data were successfully transmitted in 10-meter and 100-meter distances, measuring radiation level in a remote condition. The communication module allows an environment where radiation level can be remotely measured in an economically beneficial way as it not only consumes less electricity but also costs less. By securing linearity of a radiation measuring device and by minimizing the device itself, it is possible to set up an environment where radiation can be measured in a reliable manner, and radiation level is monitored real-time.

• Journal of Internet Computing and Services
• /
• v.22 no.1
• /
• pp.109-135
• /
• 2021

his study classified the development process of artificial intelligence (AI) speakers through analysis of the news text of artificial intelligence (AI) speakers shown in traditional news reports, and identified the characteristics of each product by period. The theoretical background used in the analysis are news frames and topic frames. As analysis methods, topic modeling and semantic network analysis using the LDA method were used. The research method was a content analysis method. From 2014 to 2019, 2710 news related to AI speakers were first collected, and secondly, topic frames were analyzed using Nodexl algorithm. The result of this study is that, first, the trend of topic frames by AI speaker provider type was different according to the characteristics of the four operators (communication service provider, online platform, OS provider, and IT device manufacturer). Specifically, online platform operators (Google, Naver, Amazon, Kakao) appeared as a frame that uses AI speakers as'search or input devices'. On the other hand, telecommunications operators (SKT, KT) showed prominent frames for IPTV, which is the parent company's flagship business, and 'auxiliary device' of the telecommunication business. Furthermore, the frame of "personalization of products and voice service" was remarkable for OS operators (MS, Apple), and the frame for IT device manufacturers (Samsung) was "Internet of Things (IoT) Integrated Intelligence System". The econd, result id that the trend of the topic frame by AI speaker development period (by year) showed a tendency to develop around AI technology in the first phase (2014-2016), and in the second phase (2017-2018), the social relationship between AI technology and users It was related to interaction, and in the third phase (2019), there was a trend of shifting from AI technology-centered to user-centered. As a result of QAP analysis, it was found that news frames by business operator and development period in AI speaker development are socially constituted by determinants of media discourse. The implication of this study was that the evolution of AI speakers was found by the characteristics of the parent company and the process of co-evolution due to interactions between users by business operator and development period. The implications of this study are that the results of this study are important indicators for predicting the future prospects of AI speakers and presenting directions accordingly.