• Journal of the Korea Society of Computer and Information
• /
• v.11 no.6 s.44
• /
• pp.279-287
• /
• 2006

This paper proposes a method to detect fall action by using stereo images to recognize emergency situation. It uses 3D information to extract the visual information for learning and testing. It uses HMM(Hidden Markov Model) as a recognition algorithm. The proposed system extracts background images from two camera images. It extracts a moving object from input video sequence by using the difference between input image and background image. After that, it finds the bounding rectangle of the moving object and extracts 3D information by using calibration data of the two cameras. We experimented to the recognition rate of fall action with the variation of rectangle width and height and that of 3D location of the rectangle center point. Experimental results show that the variation of 3D location of the center point achieves the higher recognition rate than the variation of width and height.

• Journal of Intelligence and Information Systems
• /
• v.28 no.1
• /
• pp.89-106
• /
• 2022

Recently, the amount of video data collected from smartphones, CCTVs, black boxes, and high-definition cameras has increased rapidly. According to the increasing video data, the requirements for analysis and utilization are increasing. Due to the lack of skilled manpower to analyze videos in many industries, machine learning and artificial intelligence are actively used to assist manpower. In this situation, the demand for various computer vision technologies such as object detection and tracking, action detection, emotion detection, and Re-ID also increased rapidly. However, the object detection and tracking technology has many difficulties that degrade performance, such as re-appearance after the object's departure from the video recording location, and occlusion. Accordingly, action and emotion detection models based on object detection and tracking models also have difficulties in extracting data for each object. In addition, deep learning architectures consist of various models suffer from performance degradation due to bottlenects and lack of optimization. In this study, we propose an video analysis system consists of YOLOv5 based DeepSORT object tracking model, SlowFast based action recognition model, Torchreid based Re-ID model, and AWS Rekognition which is emotion recognition service. Proposed model uses single-linkage hierarchical clustering based Re-ID and some processing method which maximize hardware throughput. It has higher accuracy than the performance of the re-identification model using simple metrics, near real-time processing performance, and prevents tracking failure due to object departure and re-emergence, occlusion, etc. By continuously linking the action and facial emotion detection results of each object to the same object, it is possible to efficiently analyze videos. The re-identification model extracts a feature vector from the bounding box of object image detected by the object tracking model for each frame, and applies the single-linkage hierarchical clustering from the past frame using the extracted feature vectors to identify the same object that failed to track. Through the above process, it is possible to re-track the same object that has failed to tracking in the case of re-appearance or occlusion after leaving the video location. As a result, action and facial emotion detection results of the newly recognized object due to the tracking fails can be linked to those of the object that appeared in the past. On the other hand, as a way to improve processing performance, we introduce Bounding Box Queue by Object and Feature Queue method that can reduce RAM memory requirements while maximizing GPU memory throughput. Also we introduce the IoF(Intersection over Face) algorithm that allows facial emotion recognized through AWS Rekognition to be linked with object tracking information. The academic significance of this study is that the two-stage re-identification model can have real-time performance even in a high-cost environment that performs action and facial emotion detection according to processing techniques without reducing the accuracy by using simple metrics to achieve real-time performance. The practical implication of this study is that in various industrial fields that require action and facial emotion detection but have many difficulties due to the fails in object tracking can analyze videos effectively through proposed model. Proposed model which has high accuracy of retrace and processing performance can be used in various fields such as intelligent monitoring, observation services and behavioral or psychological analysis services where the integration of tracking information and extracted metadata creates greate industrial and business value. In the future, in order to measure the object tracking performance more precisely, there is a need to conduct an experiment using the MOT Challenge dataset, which is data used by many international conferences. We will investigate the problem that the IoF algorithm cannot solve to develop an additional complementary algorithm. In addition, we plan to conduct additional research to apply this model to various fields' dataset related to intelligent video analysis.

• Korean Journal of Applied Biomechanics
• /
• v.26 no.2
• /
• pp.205-211
• /
• 2016

Purpose: The aim of this study was to analyze biomechanical factors and PMT (premotor time) of body muscles between elite college and amateur baseball players during the baseball batting motion. Method: Kinematic and electromyographic data were obtained for 10 elite college baseball players and 10 amateur baseball players who participated in this study. All motion capture data were collected at 200 Hz using 8 VICON cameras and the PMT of muscles was recorded using a Delsys Trigno wireless system. The peak mean bat speed and the peak mean angular velocities of trunk, pelvis, and bat with PMT of 16 body muscles were computed. These kinematic and PMT data of both groups were compared by independent t-tests (p < .05). Results: The pelvis, trunk, and bat showed a sequence of angular velocity value during baseball batting. The PMTs of right tibialis anterior, left gastrocnemius, external oblique, and erector spinae were significantly different between the two groups. Conclusion: The PMT of body muscles was related to the shifting of body and rotation of the pelvis and the trunk segment, and this action can be considered the coordinated muscle activity of the lower and upper body.

• Journal of Broadcast Engineering
• /
• v.18 no.3
• /
• pp.409-415
• /
• 2013

Recently, various equipments have been developed for convenience of presentations. Presentation equipments added the keyboard and mouse functions to laser pointer and devices have become main method. However these devices have demerits of limited action and a few events. In this paper, we propose a method which increases the degrees of freedom of presentation as the control of the hand by using a depth camera. The proposed method recognizes the horizontal and vertical positions of hand pointer and the distance between hand and camera from both depth and RGB cameras, then performs a presentation event as the location and pattern that the hand touches a screen. The simulation results show that a camera is fixed on left side of the screen, and nine presentation events is correctly performed.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2014.05a
• /
• pp.529-530
• /
• 2014

Multiplanar image video has been used in various fields advertising, exhibitions, and PR. But the content is applied therein, image synthesis, and graphics in most cases. That for taking pictures become a stock video exactly not only very difficult, to complement work in the second half of the problems of the shooting stage is difficult. Therefore, in this study, and then grasp the problems later work with imaging and attempts to validate the experiment improvements. As the study specific method, it is intended to advance the experiments for determining the distance to minimize the distortion generated at the edges of the image varies with the distance between cameras. It is intended to contribute to the creation of the content of the image-based, which can be utilized more effectively excellence media of the video if to build on this.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.3
• /
• pp.1689-1701
• /
• 2019

In recent years, personal videos have been shared online due to the popular uses of portable devices, such as smartphones and action cameras. A recent report predicted that 80% of the Internet traffic will be video content by the year 2021. Several studies have been conducted on the detection of main video events to manage a large scale of videos. These studies show fairly good performance in certain genres. However, the methods used in previous studies have difficulty in detecting events of personal video. This is because the characteristics and genres of personal videos vary widely. In a research, we found that adding a dataset with the right perspective in the study improved performance. It has also been shown that performance improves depending on how you extract keyframes from the video. we selected frame segments that can represent video considering the characteristics of this personal video. In each frame segment, object, location, food and audio features were extracted, and representative vectors were generated through a CNN-based recurrent model and a fusion module. The proposed method showed mAP 78.4% performance through experiments using LSVC data.

• Korean Educational Research Journal
• /
• v.41 no.2
• /
• pp.25-53
• /
• 2020

Purpose: School safety has recently become an important issue. In order to make a school environment safer, surveillance cameras have been installed in the facilities. The number of installations has been increasing rapidly since they have been considered widely recognizable and highly effective. However, conflicts between faculty and staff in installing and operating the system have also been increasing. In terms of school safety, these individuals'' cooperation is more than necessary. It is judged that looking into how they perceive the system could provide us with some suggestions on how to manage the related issues. The purpose of this study is to understand the perspective of faculty over closed-circuit television (CCTV) and make suggestions by analyzing their actions. Approach: In order to achieve the research objectives, I surveyed administrative actions such as the processing of official documents and CCTV policies for teachers and administrative staff of public schools in the Sejong Office of Education. In addition, I analyzed the behavior of those managing personal information and school safety-related documents along with the degree of complying with the policies. Finding: First, the correspondence rate of documents was high when there were designated document processors. Second, the acceptance level of documents in preschool was relatively low when there were designated processors. Third, the degree of accepting the policies and complying with them was higher in newly established schools than in existing schools. I found differences in the perception of how to handle the CCTV operations and the related work among the two groups of participants. In addition, I made suggestions on how to resolve the conflicts between them. Value: In this study, the education authorities quantified and measured the recognition and acceptance of faculty and staff regarding CCTV at each school level. Results showed that the active role of education authorities can make positive changes in how faculty and staff perceive the CCTV system and the problems surrounding it through school administrative action. In this regard, these results are meaningful in reducing the conflicts among the two groups and improving the organizational culture.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.709-713
• /
• 2009

Recently, a lot of research that applies data acquired from devices such as cameras and RFIDs to context aware services is being performed in the field on Life-Log and the sensor network. A variety of analytical techniques has been proposed to recognize various information from the raw data because video and audio data include a larger volume of information than other sensor data. However, manually watching a huge amount of media data again has been necessary to create supervised data for the update of a class or the addition of a new class because these techniques generally use supervised learning. Therefore, the problem was that applications were able to use only recognition function based on fixed supervised data in most cases. Then, we proposed a method of acquiring supervised data from a video sharing site where users give comments on any video scene because those sites are remarkably popular and, therefore, many comments are generated. In the first step of this method, words with a high utility value are extracted by filtering the comment about the video. Second, the set of feature data in the time series is calculated by applying functions, which extract various feature data, to media data. Finally, our learning system calculates the correlation coefficient by using the above-mentioned two kinds of data, and the correlation coefficient is stored in the DB of the system. Various other applications contain a recognition function that is used to generate collective intelligence based on Web comments, by applying this correlation coefficient to new media data. In addition, flexible recognition that adjusts to a new object becomes possible by regularly acquiring and learning both media data and comments from a video sharing site while reducing work by manual operation. As a result, recognition of not only the name of the seen object but also indirect information, e.g. the impression or the action toward the object, was enabled.

• Korean Journal of Applied Biomechanics
• /
• v.26 no.1
• /
• pp.71-82
• /
• 2016

Objective: The purpose of this study was to investigate kinematic comparisons of Tsukahara vault in gymnastics between a top-level athlete and sublevel collegiate athletes in order to obtain information on key biomechanical points for successful Tsukahara vaults. Methods: An Olympic gold medalist (height, 160 cm; weight, 52 kg; age, 25 years) and five sublevel collegiate gymnasts (height, $168.2{\pm}3.4cm$; weight, $59.6{\pm}3.1kg$; age, $23.2{\pm}1.6years$) participated in this study. They repeatedly performed Tsukahara vaults including one somersault. Fourteen motion-capturing cameras were used to collect the trajectories of 26 body markers during Tsukahara vaults. Event time, displacement and velocity of the center of mass, joint angles, the distance between the two hands on the horse, and averaged horizontal and vertical impact forces were calculated and compared. Results: The top-level athlete showed a larger range of motion (ROM) of the hip and knee joints compared to sublevel collegiate athletes during board contact. During horse contact, the top-level athlete had a narrow distance between the two hands with extended elbows and shoulders in order to produce a strong blocking force from the horse with a shorter contact time. At the moment of horse take-off, reactive hip extension of the top-level athlete enhanced propulsive take-off velocity and hip posture during post-flight phase. Conclusion: Even though a high velocity of the center of mass is important, the posture and interactive action during horse contact is crucial to post-flight performance and the advanced performance of Tsukahara vaults.

• Smart Media Journal
• /
• v.1 no.2
• /
• pp.12-22
• /
• 2012

Today we are witnessing an increasingly widespread use of cameras in our lives for video surveillance, robot vision, and mobile phones. This has led to a renewed interest in computer vision in general and an on-going boom in human activity recognition in particular. Although not particularly fancy per se, human gait is inarguably the most common and frequent action. Early on this decade there has been a passing interest in human gait recognition, but it soon declined before we came up with a systematic analysis and understanding of walking motion. This paper presents a set of DBN-based models for the analysis of human gait in sequence of increasing complexity and modeling power. The discussion centers around HMM-based statistical methods capable of modeling the variability and incompleteness of input video signals. Finally a novel idea of extending the discrete state Markov chain with a continuous density function is proposed in order to better characterize the gait direction. The proposed modeling framework allows us to recognize pedestrian up to 91.67% and to elegantly decode out two independent gait components of direction and posture through a sequence of experiments.