• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.60 no.8
• /
• pp.1577-1585
• /
• 2011

This paper proposes a posture stabilization control algorithm for a wheeled mobile robot using hybrid visual servo control method with a position based and an image based visual servoing (PBVS and IBVS). To overcome chattering phenomena which were shown in the previous researches using a simple switching function based on a threshold, the proposed hybrid visual servo control law introduces the fusion function based on a blending function. Then, the chattering problem and rapid motion of the mobile robot can be eliminated. Also, we consider the nonlinearity of the wheeled mobile robot unlike the previous visual servo control laws using linear control methods to improve the performances of the visual servo control law. The proposed posture stabilization control law using hybrid visual servoing is verified by a theoretical analysis and simulation and experimental results.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.67 no.7
• /
• pp.934-939
• /
• 2018

Recently, there have been a lot of sleepy driving accidents. In this study, we conducted a preliminary study for detecting drowsiness using posture and image processing technology. We used pressure sensors to study posture. We also investigated the possibility of drowsy recognition using histogram. As a result of the experiment, it was possible to distinguish positions through pressure sensors. Also, it was confirmed that the drowsiness phenomenon can be distinguished by using the histogram.

• The Research Journal of the Costume Culture
• /
• v.25 no.4
• /
• pp.529-542
• /
• 2017

The purpose of this study is to utilize technology as basic data for smart clothing product research and development. This technology can recognize user's motion according to characteristics types and functions of wearable smart clothing products. In order to analyze the case of motion recognition products, we searched for previous research data and cases referred to as major keywords in leading search engines, Google and Naver. Among the searched cases, information on the characteristics and major functions of the 42 final products selected on the market are examined in detail. Motion recognition for smart clothing products is classified into four body types: head & face, body, arms & hands, and legs & feet. Smart clothing products was developed with various items, such as hats, glasses, bras, shirts, pants, bracelets, rings, socks, shoes, etc., It was divided into four functions health care type for prevention of injuries, health monitor, posture correction, sports type for heartbeat and exercise monitor, exercise coaching, posture correction, convenience for smart controller and security and entertainment type for pleasure. The function of the motion recognition smart clothing product discussed in this study will be a useful reference when designing a motion recognition smart clothing product that is blended with IT technology.

• Journal of Digital Convergence
• /
• v.18 no.12
• /
• pp.97-103
• /
• 2020

The proposed system is composed of two parts, an AI physical fitness measurement part and an AI physical fitness management part. In the AI fitness measurement part, a guide to physical fitness measurement and accurate calculation of the measured value are performed through deep learning-based pose recognition. Based on these measurements, the AI fitness management part designs personalized exercise programs and provides them to dedicated smart applications. To guide the measurement posture, the posture of the subject to be measured is photographed through a webcam and the skeleton line is extracted. Next, the skeletal line of the learned preparation posture is compared with the extracted skeletal line to determine whether or not it is normal, and voice guidance is provided to maintain the normal posture.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.8
• /
• pp.1925-1936
• /
• 2010

In this paper, multi-legged robot was designed and produced using stable walking pattern algorithm. The robot had embedded camera and wireless communication function and it is possible to recognize both hand posture and obstacles. The algorithm decided moving paths, and recognized and avoided obstacles through Hough Transform using Edge Detection of inputed image from image sensor. The robot can be controlled by hand posture using Mahalanobis Distance and average value of skin's color pixel, which is previously learned in order to decide the destination. The developed system has shown obstacle detection rate of 96% and hand posture recognition rate of 94%.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.298-300
• /
• 2021

The Ministry of National Defense decided to build a realistic combat simulation training system based on virtual reality and augmented reality in accordance with the expansion of the scientific training system of "Defense Reform 2.0". The realistic combat simulation training system should be able to maximize the tension and training effect as in actual combat through engagement between trainers. In addition, it should be possible to increase the effectiveness of survival training at the same time as shooting training similar to actual combat through cover training. Previous studies are suitable techniques to improve the shooting precision of the trainee, but it is difficult to practice bilateral engagement like in actual combat, and it is particularly insufficient for combat shooting training using cover. Therefore, in this paper, we propose a S/W algorithm for generating a virtual avatar by recognizing the shooting posture of the opponent on the screen of the virtual shooting trainer. This S/W algorithm can recognize the trainer and the cover based on the depth information acquired through the depth sensor and estimate the trainer's posture.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.5
• /
• pp.177-183
• /
• 2022

Recently, interesting of home training is getting bigger due to COVID-19. Accordingly, research on applying HAR(human activity recognition) technology to home training has been conducted. However, existing paper of HAR proposed static activity instead of dynamic activity. In this paper, the deep learning model where dynamic exercise posture can be analyzed and the accuracy of the user's exercise posture can be shown is proposed. Fitness images of AI-hub are analyzed by blaze pose. The experiment is compared with three types of deep learning model: RNN(recurrent neural network), LSTM(long short-term memory), CNN(convolution neural network). In simulation results, it was shown that the f1-score of RNN, LSTM and CNN is 0.49, 0.87 and 0.98, respectively. It was confirmed that CNN is more suitable for human activity recognition than other models from simulation results. More exercise postures can be analyzed using a variety learning data.

• Journal of the Korean Society of Industry Convergence
• /
• v.26 no.1
• /
• pp.113-119
• /
• 2023

In this paper, we propose an optimization of a pose estimation deep learning model for recognition of abnormal behavior in unmanned stores using radio frequencies. The radio frequency use millimeter wave in the 30 GHz to 300 GHz band. Due to the short wavelength and strong straightness, it is a frequency with less grayness and less interference due to radio absorption on the object. A millimeter wave radar is used to solve the problem of personal information infringement that may occur in conventional CCTV image-based pose estimation. Deep learning-based pose estimation models generally use convolution neural networks. The convolution neural network is a combination of convolution layers and pooling layers of different types, and there are many cases of convolution filter size, number, and convolution operations, and more cases of combining components. Therefore, it is difficult to find the structure and components of the optimal posture estimation model for input data. Compared with conventional millimeter wave-based posture estimation studies, it is possible to explore the structure and components of the optimal posture estimation model for input data using genetic algorithms, and the performance of optimizing the proposed posture estimation model is excellent. Data are collected for actual unmanned stores, and point cloud data and three-dimensional keypoint information of Kinect Azure are collected using millimeter wave radar for collapse and property damage occurring in unmanned stores. As a result of the experiment, it was confirmed that the error was moored compared to the conventional posture estimation model.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.22 no.1
• /
• pp.115-125
• /
• 2023

This paper relates to the development of CCTV camera posture calibration algorithm that can simultaneously collect traffic information such as traffic volume and speed in the state of view of the CCTV camera set for traffic monitoring. The developed autonomous posture calibration algorithm uses vehicle recognition and tracking techniques to identify the road, and automatically determines the angle of view for the operator's traffic surveillance and traffic information collection. To verify the performance of the proposed algorithm, a CCTV installed on site was used, and the results of the angle of view automatically calculated by the autonomous posture calibration algorithm for the angle of view set for traffic surveillance and traffic information collection were compared.

• Journal of Korea Multimedia Society
• /
• v.24 no.5
• /
• pp.659-666
• /
• 2021

With the continuous development of deep learning, human behavior recognition algorithms have achieved good results. However, in a multi-person recognition environment, the complex behavior environment poses a great challenge to the efficiency of recognition. To this end, this paper proposes a multi-person pose estimation model. First of all, the human detectors in the top-down framework mostly use the two-stage target detection model, which runs slow down. The single-stage YOLOv3 target detection model is used to effectively improve the running speed and the generalization of the model. Depth separable convolution, which further improves the speed of target detection and improves the model's ability to extract target proposed regions; Secondly, based on the feature pyramid network combined with context semantic information in the pose estimation model, the OHEM algorithm is used to solve difficult key point detection problems, and the accuracy of multi-person pose estimation is improved; Finally, the Euclidean distance is used to calculate the spatial distance between key points, to determine the similarity of postures in the frame, and to eliminate redundant postures.