• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.10
• /
• pp.1262-1270
• /
• 2020

Recently, the smart video surveillance research, which has been receiving increasing attention, has mainly focused on the intruder detection and tracking, and abandoned object detection. On the other hand, research on real-time detection of stolen objects is relatively insufficient compared to its importance. Considering various smart surveillance video application environments, this paper presents two different types of stolen object detection algorithms. We first propose an algorithm that detects theft of statically and dynamically registered surveillance objects using a dual background subtraction model. In addition, we propose another algorithm that detects theft of general surveillance objects by applying the dual background subtraction model and Mask R-CNN-based object segmentation technology. The former algorithm can provide economical theft detection service for pre-registered surveillance objects in low computational power environments, and the latter algorithm can be applied to the theft detection of a wider range of general surveillance objects in environments capable of providing sufficient computational power.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.19 no.6
• /
• pp.61-73
• /
• 2020

Real-time location-sensing technology using location information collected from IoT devices is being applied for safety management purposes in many industries, such as ports. On the other hand, positional error is always present owing to the characteristics of GPS. Therefore, accident-risk detection algorithms must consider positional error. This paper proposes an methodology of calibration for falling object accident-risk-zone approach detection algorithm considering GPS errors. A probability density function was estimated, with positional error data collected from IoT devices as a probability variable. As a result of the verification, the algorithm showed a detection accuracy of 93% and 77%. Overall, the analysis results derived according to the GPS error level will be an important criterion for upgrading algorithms and real-time risk managements in the future.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.9
• /
• pp.139-147
• /
• 2022

The goal of this study is to develop a mobile application so that a person who is new to chemistry can easily acquire the knowledge necessary for chemical structure learning using image tracking technology. The point of this study is to provide a new chemical structure learning experience by recognizing a two-dimensional picture, augmenting the chemical structure into a three-dimensional object, showing it on the user's screen, and using a service that simultaneously provides related information in multiple fields. characteristic. Login API and real-time database technology were used for safe and real-time data management, and an application was developed using image tracking technology for image recognition and 3D object augmentation service. In the future, we plan to use the chemical structure data library to efficiently load and output data.

• Journal of Digital Policy
• /
• v.1 no.1
• /
• pp.15-20
• /
• 2022

Recently, the healthcare field is trying to develop a model that can improve service quality by reflecting the requirements of various industrial fields. In this paper, we propose an Internet of Behavior (IoB) environment model that can process users' healthcare information in real time in a 5G environment to improve healthcare services. The purpose of the proposed model is to analyze the user's healthcare information through deep learning and then check the health status in real time. In this case, the biometric information of the user is transmitted through communication equipment attached to the portable medical equipment, and user authentication is performed through information previously input to the attached IoB device. The difference from the existing IoT healthcare service is that it analyzes the user's habits and behavior patterns and converts them into digital data, and it can induce user-specific behaviors to improve the user's healthcare service based on the collected data.

• Smart Media Journal
• /
• v.12 no.7
• /
• pp.59-67
• /
• 2023

Recently, research on behavior analysis tracking human posture and movement has been actively conducted. In particular, OpenPose, an open-source software developed by CMU in 2017, is a representative method for estimating human appearance and behavior. OpenPose can detect and estimate various body parts of a person, such as height, face, and hands in real-time, making it applicable to various fields such as smart healthcare, exercise training, security systems, and medical fields. In this paper, we propose a method for classifying four exercise movements - Squat, Walk, Wave, and Fall-down - which are most commonly performed by users in the gym, using OpenPose-based deep learning models, DNN and CNN. The training data is collected by capturing the user's movements through recorded videos and real-time camera captures. The collected dataset undergoes preprocessing using OpenPose. The preprocessed dataset is then used to train the proposed DNN and CNN models for exercise movement classification. The performance errors of the proposed models are evaluated using MSE, RMSE, and MAE. The performance evaluation results showed that the proposed DNN model outperformed the proposed CNN model.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.3
• /
• pp.150-159
• /
• 2022

This paper presents an implementation of detecting whether a helmet is worn and there is a fall accident through individual image analysis in real-time from extracting the image objects of several workers active in the industrial field. In order to detect image objects of workers, YOLO, a deep learning-based computer vision model, was used, and for whether a helmet is worn or not, the extracted images with 5,000 different helmet learning data images were applied. For whether a fall accident occurred, the position of the head was checked using the Pose real-time body tracking algorithm of Mediapipe, and the movement speed was calculated to determine whether the person fell. In addition, to give reliability to the result of a falling accident, a method to infer the posture of an object by obtaining the size of YOLO's bounding box was proposed and implemented. Finally, Telegram API Bot and Firebase DB server were implemented for notification service to administrators.

• Journal of the Korea Computer Graphics Society
• /
• v.29 no.3
• /
• pp.149-158
• /
• 2023

HMD, which is applied with the latest technology, minimizes motion sickness with high-resolution displays and fast motion recognition, and can accurately track location and motion. This can provide an environment where you can immerse yourself in a virtual three-dimensional space, and virtual reality contents such as disaster simulators and high-risk equipment learning spaces are developing using these characteristics. These advantages are also applicable in the field of basic science education. In particular, expanding the concepts of electric and magnetic fields in physics described by existing two-dimensional data into three-dimensional spaces and visualizing them in real time can greatly help improve learning understanding. In this paper, realistic physical education environments and contents based on three-dimensional virtual reality are developed and the developed learning contents are experienced by actual learning subjects to prove their effectiveness. A total of 46 middle school and college students were taught and experienced in real time the electric and magnetic fields expressed in three dimensions in a virtual reality environment. As a result of the survey, more than 85% of positive responses were obtained, and positive results were obtained that three-dimensional virtual space-based physical learning could be effectively applied.

• Annual Conference of KIPS
• /
• 2022.11a
• /
• pp.565-567
• /
• 2022

질병을 앓고 있는 환자는 상태에 따라 병실, 주거지, 요양원 등 조호환경 내 생활 시 의료 인력의 지속적인 추적 및 관찰을 통해 신체에 이상이 생긴 경우 이를 감지하고, 신속하게 조치할 수 있도록 해야 한다. 의료 인력이 직접 환자를 확인하는 방법은 의료 인력의 반복적인 노동이 요구되며 실시간으로 환자를 확인해야 한다는 특성상 의료 인력이 상주해야 하기에 이는 곧, 의료 인력의 부족과 낭비로 이어진다. 해당 문제 해결을 위해 의료 인력을 대신하여 조호환경 내 환자의 상태를 실시간으로 모니터링할 수 있는 딥러닝 모델들이 연구되고 있다. 딥러닝 모델은 데이터의 수가 많을수록 강인한 모델을 설계할 수 있으며, 데이터셋의 배경, 객체의 특징 분포 등 다양한 조건에 영향을 받기 때문에 학습에 필요한 도메인을 가지는 많은 양의 전처리된 데이터를 수집해야 한다. 따라서, 조호환경 내 환자에 대한 데이터셋이 필요하지만, 공개된 데이터셋의 경우 양이 매우 적으며 이를 반전, 회전기법 등을이용할 경우 데이터의 수를 늘릴 수 있지만, 같은 분포의 특징을 가지는 데이터가 생성되기에 데이터 증강 기법을 단순하게 적용하면 딥러닝 모델의 과적합을 야기한다. 또한, 조호환경 내 이미지 데이터셋은 얼굴 노출과 같은 개인정보가 포함 될 수 있으며 이를 보호하기 위해 정보들을 비식별화 해야 한다는 문제점이 있다. 따라서 본 논문에서는 조호환경에서 수집된 데이터 증강을 위한 Style-Generative Adversarial Networks 기법을 적용하여 조호환경 데이터셋 수집에 효과적인 증강 기법을 제안한다.

• 전자공학회논문지 IE
• /
• v.47 no.4
• /
• pp.18-27
• /
• 2010

This paper is designed the efficient Data Acquisition System for an vibration of rotatory machines. The Data Acquisition System is consist of the analog logic having signal filer and amplifier, and digital logic with ADC, DSP, FPGA and FIFO memory. The vibration signal of rotatory machines acquired from sensors is controlled by the FPGA device through the analog logic and is saved to FIFO memory being converted analog to digital signal. The digital signal process is performed by the DSP using the vibration data in FIFO memory. The vibration factor of the rotatory machinery analysis and diagnosis is defined the RMS, Peak to Peak, average, GAP, FFT of vibration data and digital filtering by DSP, and is need to follow as being happened the event of vibration and make an application to an warning system. It takes time to process the several analysis step of all vibration data and the event follow, also special event. It should be continuously performed the data acquisition and the process, however during processing the input signal the DSP can not be performed to the acquisited data after then, also it will be lose the data at several channel. Therefore it is that the system uses efficiently the DSP and FPGA devices for reducing the data lose, it design to process a part of the signal data to FPGA from DSP in order to minimize the process time, and a process to parallel process system, as a result of design system it propose to method of faster process and more efficient data acquisition system by using DSP and FPGA than signal DSP system.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.9
• /
• pp.661-674
• /
• 2009

Recently, there has been much research and many attempts to enhance converged information technology services using new technology such as ubiquitous sensor networks (USN) in medical, environmental, industrial, and logistic areas. There has also been much research and various attempts to apply this new technology to agricultural areas. However, applications to the agricultural areas should be considered differently against the same areas such as medical, environmental, industrial, and logistics. Therefore, this paper suggests that an agricultural cultivating management and traceability system. This system is a unified system that supports the processes sowing seeds through selling agricultural products to consumers. Farmers can be provided with an effective calendar for cultivation and weather information in real time as well as the monitoring of the growth of farm products on the farm in real time using this system. Farmers can also control all equipment installed on the farm directly or remotely and the equipment can be controlled automatically when the measured values such as temperature and humidity deviate from the decent criteria which are set by farmers or this system. Additionally, the reliability for and the better quality of the agricultural products can be improved because farmers can use this unified system to cover all processes from sowing seeds to selling to consumers.