• 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.

• Proceedings of the Acoustical Society of Korea Conference
• /
• autumn
• /
• pp.97-100
• /
• 1999

본 논문은 자동차의 편의성 및 안전성의 동시 확보를 위하여, 보조적 스위치의 조작없이 상시 음성의 입$\cdot$출력이 가능하도록 하였고, 남성과 여성을 구별하기 위하여 피치검출법을 사용하여 속도별로 구분하였다. 또한, band pass filter를 이용하여 자동으로 잡음하에서 정확하게 음성추간 검출(End Point Detection)을 하게 하였다. Reference Pattern은 DMS(Dynaminc Multi-Section)[1]모델을 사용하려고, 음성의 특징 파라미터와 인식 알고리즘은 PLP 13차와 One Stage Dynamic Programming(OSDP)를 사용하였다. 시내주행중인 자동차 환경에서 자주 사용되는 차량제어 명령어 30단어를 가지고 실험한 결과 40-80km에서 화자독립 남성 $96\%$, 여성 $94.4\%$ 화자종속일 때 남성 $97\%$, 여성 $95\%$의 인식률을 얻을수 있었고 남성과 여성을 구분하므로 써 인식률을 향상 시켰다.

• Bulletin of the Korean Chemical Society
• /
• v.16 no.7
• /
• pp.595-600
• /
• 1995

The kinetics of the reaction between O(3P) and C2H4 was investigated by measuring time-dependent concentrations of OH resulting from the reaction by using the LIF detection. Oxygen atoms were generated by titrating microwave discharged N2/He with NO to the chemiluminescent end point. The operating pressures in the flow reactor ranged from 5 to 15 torr and the mixtures consisted of He/O(3/P)/C2H4 in the approximate ratios from 100/1/0.1 to 100/1/1. The controlled residence time prior to the detection were estimated to be 0.8-17 ms at the reactor pressure of 7 torr. Experimentally determined profiles both in shape and magnitude were compared with the computed OH density for a specified set of experimental parameters, allowing us to arrive at a complete mechanism for the reaction of O(3P) with ethylene.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.31 no.4
• /
• pp.661-674
• /
• 2021

Malware, such as a rootkit that modifies the kernel, can adversely affect the analyst's judgment, making the analysis difficult or impossible if a mechanism to evade memory analysis is added. Therefore, we plan to preemptively respond to malware such as rootkits that bypass detection through advanced kernel modulation in the future. To this end, the main structure used in the Windows kernel was analyzed from the attacker's point of view, and a method capable of modulating the kernel object was applied to modulate the memory dump file. The result of tampering is confirmed through experimentation that it cannot be detected by memory analysis tool widely used worldwide. Then, from the analyst's point of view, using the concept of tamper resistance, it is made in the form of software that can detect tampering and shows that it is possible to detect areas that are not detected by existing memory analysis tools. Through this study, it is judged that it is meaningful in that it preemptively attempted to modulate the kernel area and derived insights to enable precise analysis. However, there is a limitation in that the necessary detection rules need to be manually created in software implementation for precise analysis.

• Proceedings of the KIEE Conference
• /
• 2003.11a
• /
• pp.88-90
• /
• 2003

In this paper, an algorithm to compensate the distorted signals due to CT(Current Transformer) saturation is suggested. Firstly, WT(Wavelet Transform) is used to detect a start point and an end point of saturation. Filter banks which can be easily realized in real-time applications are employed in detecting CT saturation. Secondly, least-square curve fitting method is used to restore the distorted section of the secondary current. Fault simulations are performed on a power system model using EMTP(Electromagnetic Transient Program). A series of test results indicate that WT has superior detection accuracy and the proposed algorithm which shows very stable features under various levels of remanent flux is also satisfactory.

• Journal of the Korean Chemical Society
• /
• v.11 no.4
• /
• pp.159-164
• /
• 1967

An exact expression of the titration fraction as a function of the potential is derived for the cases where the coefficients of the both half reactions involved in the titration are homogeneous. It shows that the potential is independent of the concentration of the reagents not only at the equivalence point but also at all titration fractions. The sharpness of the end point detection by potentiometric method is shown to depend not only on the difference of the normal potentials involved but also strongly on the number of electrons transferred in each half reaction. The inflexion point of the potentiometric titration curve is shown to be slightly off from the equivalence point, including the cases where the number of electrons involved are equal. Completeness of the reaction in the course of titration is analyzed, too, mostly in terms of equilibrium constant, thus most of the results are applicable to any type of equilibrium in a single phase with particular relationship of coefficients of chemical equation.

• Journal of Digital Contents Society
• /
• v.18 no.5
• /
• pp.989-994
• /
• 2017

In this paper, we propose a pothole detection system using 2D LiDAR and a pothole detection algorithm. Conventional pothole detection methods can be divided into vibration-based method, 3D reconstruction method, and vision-based method. Proposed pothole detection system uses two inexpensive 2D LiDARs and improves pothole detection performance. Pothole detection algorithm is divided into preprocessing for noise reduction, clustering and line extraction for visualization, and gradient function for pothole decision. By using gradient of distance data function, we check the existence of a pothole and measure the depth and width of the pothole. The pothole detection system is developed using two LiDARs, and the 3D pothole detection performance is shown by detecting a pothole with moving LiDAR system.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.17 no.2
• /
• pp.152-167
• /
• 2018

The high-speed weigh-in-motion system can collect the traveling speed and load information of individual vehicles, which can be used in a variety of ways for the traffic surveillance. However, it has a limit to apply the high-speed weigh-in-motion data directly to a safety analysis because high-speed weigh-in-motion's raw data are point measured data. In order to overcome this problem, this paper proposes a method to calculate the conflict rate and the Impulse severity based on surrogate safety measures derived from the detection time, detection speed, vehicle length, vehicle type, vehicle weight. It will be possible to analyze and evaluate the risk of rear-end collision on freeway traffic. In addition, this study is expected to be used as a fundamental for identifying crash risks and developing policies to enhance traffic safety on freeways.

• Korean Journal of Clinical Laboratory Science
• /
• v.48 no.2
• /
• pp.124-129
• /
• 2016

Endotoxin, also known as lipopolysaccharide (LPS) produced by the cell wall of gram negative bacteria can be present in any liquid or on any biomaterial. Endotoxin in blood can cause fever and inflammation. In this study, we compared bacterial endotoxin using Escherichia coli O157:H7, Klebsiella oxytoca, Salmonella Typhi, Shigella sonnei and Morganella morganii. Bacteria were cultured for use in the experiment, and diluted to $1.5{\times}10^8CFU/mL$. A check marked sensitivity confirmatory test of the Limulus amebocyte lysate (LAL) reagent was performed to examine the validity. The end point reaction to each bacteria sample was confirmed with 10 fold dilution and then the final reaction end point was confirmed by 2 fold dilution between the dilution step and the upper dilution step. According to the results, in detection of endotoxins in more than 0.015 EU/mL, E. coli O157 was 75~37.5 CFU/mL, K. oxytoca 37.5~18.75 CFU/mL, M. morganii and S. Typhi 3.75~1.875 CFU/mL, and S. sonnei 7.5~3.75 CFU/mL. The resulting value was finally ensured by a confirmation test for the inhibitory factor. Based on this study, conduct of further research on bacterial endotoxin is encouraged.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2001.11b
• /
• pp.181-184
• /
• 2001

Chemical Mechanical Polishing(CMP) of Shallow Trench Isolation(STD structure in 0.18 m semiconductor device fabrication is studied. CMP process is applied for the STI structure with and without reverse moat pattern and End Point Detection (EPD) method is tested. To optimize the transistor properties related metal 1 parameters. we studied the correlation between CMP thickness of STI using high selectivity slurry. DOE of gate etch recipe, and 1st metal DC values. Remaining thickness of STI CMP is proportional to the thickness of gate-etch process and this can affect to gate profile. As CMP thickness increased. the N-poly foot is deteriorated. and the P-Poly Noth is getting better. If CD (Critical Dimension) value is fixed at some point,, all IDSN/P values are in inverse proportional to CMP thickness by reason of so called Profile Effect. Weve found out this phenomenon in all around DOE conditions of Gate etch process and we also could understand that it would not have any correlation effects between VT and CMP thickness in the range of POE 120 sec conditions. As CMP thickness increased by $100\AA$. 3.2 $u\AA$ of IDSN is getting better in base 1 condition. In POE 50% condition. 1.7 $u\AA$ is improved. and 0.7 $u\AA$ is improved in step 2 condition. Wed like to set the control target of CD (critical dimension) in gate etch process which can affect Idsat, VT property versus STI topology decided by CMP thickness. We also would like to decide optimized thickness target of STI CMP throughout property comparison between conventional STI CMP with reverse moat process and newly introduced STI CMP using high selectivity slurry. And we studied the process conditions to reduce Gate Profile Skew of which source known as STI topology by evaluation of gate etch recipe versus STI CMP thickness.