• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.31-34
• /
• 2018

대화 시스템에서 사람과 기계와의 모든 발화에서 발생하는 상황들을 모두 규칙화할 수 없기 때문에 자연스러운 대화가 단절되는 breakdown 현상이 빈번하게 일어날 수 있다. 이런 현상이 발생하는 이유는 다음과 같다. 첫째, 대화에서는 다양한 도메인이 등장하기 때문에 시스템이 커버할 수 있는 리소스가 부족하며, 둘째, 대화 데이터에서 학습을 위한 annotation되어 있는 많은 양의 코퍼스를 보유하기에는 한계가 있으며, 모델에 모든 대화 흐름의 히스토리를 반영하기 어렵다. 이런 한계점이 존재함에도 breakdown detection은 자연스러운 대화 시스템을 위해서는 필수적인 기능이다. 본 논문은 이런 이슈들을 해소하기 위해서 memory attention기반의 새로운 모델을 제안하였다. 제안한 모델은 대화내에 발화에 대해 memory attention을 이용하여 과거 히스토리가 반영되기 때문에 자연스러운 대화흐름을 잘 detection할 수 있으며, 기존 모델과의 성능비교에서 state-of-the art 결과를 도출하였다.

• ETRI Journal
• /
• v.33 no.3
• /
• pp.415-422
• /
• 2011

In this paper, we present a novel vision-based method of recognizing finger actions for use in electronic appliance interfaces. Human skin is first detected by color and consecutive motion information. Then, fingertips are detected by a novel scale-invariant angle detection based on a variable k-cosine. Fingertip tracking is implemented by detected region-based tracking. By analyzing the contour of the tracked fingertip, fingertip parameters, such as position, thickness, and direction, are calculated. Finger actions, such as moving, clicking, and pointing, are recognized by analyzing these fingertip parameters. Experimental results show that the proposed angle detection can correctly detect fingertips, and that the recognized actions can be used for the interface with electronic appliances.

• The Transactions of the Korean Institute of Electrical Engineers A
• /
• v.48 no.11
• /
• pp.1472-1475
• /
• 1999

This paper proposes a new detection method of human faces in grey scale images with cluttered background using a facial template and elliptical structure of the human head. Face detection technique can be applied in many areas of image processing such as face recognition, composition and computer graphics, etc. Until now, many researches about face detection have been done, and applications in more complicated conditions are increasing. The existing technique proposed by Sirohey shows relatively good performance in image with cluttered background, but can apply only to image with one face and needs much computation time. The proposed method is designed to reduce complexity and be applied even in the image with several faces by introducing template matching as preprocess. The results show that the proposed method produces more correct detection rate and needs less computation time than the existing one.

• Journal of Navigation and Port Research
• /
• v.32 no.2
• /
• pp.133-140
• /
• 2008

In this paper, a fault detection method for operator failures using the observation technique is proposed. The suggested algorithm is extended using the conventional sensor/actuator fault detection method. First, it is assumed that operator failure affects human work operations, as it is an external input signal. With this assumption, a human work model with operator failure is suggested. Second, an unknown input observer with proportional and integral gains is introduced. The characteristic of this observer of estimating an external signal without an exact input is shown, and the conditions for the detection of an operator failure are proposed. Finally, by simulating the container crane operations, it is verified that the observer can accurately detect an operator failure and estimate its magnitude from the given internal signal.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.10 no.3
• /
• pp.333-340
• /
• 2015

In this paper, we propose a passive microwave sensor for detecting human body using microwave radiometer technologies. The proposed sensor detects human body by measuring the change of the received radiation power from fixed background object due to human body. A C-band microwave radiometer is designed and implemented. The received radiation power changes due to human body is measured by the C-band microwave radiometer, and the effectiveness of the proposed sensor is evaluated by the measurement result analysis.

• Proceedings of the IEEK Conference
• /
• 1999.06a
• /
• pp.421-424
• /
• 1999

Automatic human face detection in a complex background is one of the difficult problems In this paper. we propose an effective automatic face detection system that can locate the face region in natural scene images when the system is used as a pre-processor of a face recog- nition system. We use two natural and powerful visual cues, the color and the human head shape. The outline of the human head can be generally described as being roughly elliptic in nature. In the first step of the proposed system, we have tried the approach of fitting the best Possible ellipse to the outline of the head In the next step, the method based on the human skin color space by selecting flesh tone regions in color images and histogramming their r(＝R/(R＋G＋B)) and g(＝G/R＋G＋B)) values. According to our experiment. the proposed system shows robust location results

• Journal of Korea Multimedia Society
• /
• v.21 no.4
• /
• pp.478-484
• /
• 2018

The choice of the motion features influences the result of the human action recognition method directly. Many factors often influence the single feature differently, such as appearance of the human body, environment and video camera. So the accuracy of action recognition is restricted. On the bases of studying the representation and recognition of human actions, and giving fully consideration to the advantages and disadvantages of different features, the Dense Trajectories(DT) algorithm is a very classic algorithm in the field of behavior recognition feature extraction, but there are some defects in the use of optical flow images. In this paper, we will use the improved Dense Trajectories(iDT) algorithm to optimize and extract the optical flow features in the movement of human action, then we will combined with Support Vector Machine methods to identify human behavior, and use the image in the KTH database for training and testing.

• Journal of Korea Multimedia Society
• /
• v.21 no.12
• /
• pp.1459-1466
• /
• 2018

Robots which detect human and recognize action are important factors for human interaction, and many researches have been conducted. Recently, deep learning technology has developed and learning based robot's technology is a major research area. These studies require a database to learn and evaluate for intelligent human perception. In this paper, we propose a multi-modal sensor-based image database condition considering the security task by analyzing the image database to detect the person in the outdoor environment and to recognize the behavior during the running of the robot.

• Proceedings of the KACD Conference
• /
• 2003.11a
• /
• pp.546.1-546
• /
• 2003

Laser doppler flowmeter (LDF) has been applied to the measurement of pulpal blood flow (PBF) in human teeth. As far as we searched, the detection area of the pulp in the blood flow measurement has not been clarified, yet. Therefore, the purpose of this study was to obtain information of the detection area in PBF measurement using LDF. The experiments were performed on the artificial blood circulation in extracted human upper central incisors. The apical portions of examined teeth (n=6) were severed and root canals were enlarged from the apical end to the 2mm incisal to the level of enamel-cement junction. An individual resin cap of each tooth was prepared and a hole was drilled 2mm incisal to enamel-cement junction of the labial side of the cap. The measurement probe of LDF (MBF3D, Moor Instrument, UK) was plugged into the hole of the cap. Heparinized human peripheral blood, which was in advance collected and diluted 3 times with physiological saline, was pumped through the apical foramen of the teeth via a silicone tube and a disposable needle (o.d. 0.7mm) and blood flow signals were monitored. The flux signal significantly increased with the enlargement of the root canal to incisal direction (p<0.01, Friedman analysis). The result indicates that the performance of LDF in PBF with human teeth is limited.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.65 no.5
• /
• pp.836-842
• /
• 2016

When an accident occurs in the nuclear power plant, the faulted information might mislead to the high possibility of aggravating the accident. At the Fukushima accident, the operators misunderstood that there was no core exposure despite in the processing of core damage, because the instrument information of the reactor water level was provided to the operators optimistically other than the actual situation. Thus, this misunderstanding actually caused to much confusions on the rapid countermeasure on the accident, and then resulted in multiplying the accident propagation. It is necessary to be equipped with the function that informs operators the status of instrument integrity in real time. If plant operators verify that the instruments are working properly during accident conditions, they are able to make a decision more safely. In this study, we have performed various tests for the fault detection sensitivity of an data-driven empirical model to review the usability of the model in the accident conditions. The test was performed by using simulation data from the compact nuclear simulator that is numerically simulated to PWR type nuclear power plant. As a result of the test, the proposed model has shown good performance for detecting the specified instrument faults during normal plant conditions. Although the instrument fault detection sensitivity during plant accident conditions is lower than that during normal condition, the data-drive empirical model can be detected an instrument fault during early stage of plant accidents.