• Science of Emotion and Sensibility
• /
• v.15 no.1
• /
• pp.1-8
• /
• 2012

There have been many emotion researches to investigate physiological responses on specific emotions with physiological parameters such as heart rate, blood volume flow, and skin conductance. Very few researches, however, exists by detecting them with facial skin temperature. The purpose of present study was to observe the differences of facial skin temperature by using thermal camera, when participants stimulated by monitor scenes which could evoke fear or joy. There were totally 98 of participants; undergraduate students who were in their adult age and middle, high school students who were in their adolescence. We measured their facial temperature, before and after presenting emotional stimulus to see changes between both times. Temperature values were extracted in these regions; forehead, inner corners of the eyes, bridge of the nose, end of the nose, and cheeks. Temperature values in bridge and end of the nose were significantly decreased in fear emotion stimulated. There was also significant temperature increase in the area of forehead and the inner corners of the eyes, while the temperature value in end of the nose decreased. It showed decrease in both stimulated fear and joy. These results might be described as follows: When arousal level going up, sympathetic nervous activity increases, and in turn it makes blood flow in peripheral vessels under the nose decrease. Facial temperature changes by fear or joy in this study were the same as the previous studies which measured temperature of finger tip, when participants experiencing emotions. Our results may help to develop emotion-measuring techniques and establish computer system bases which are to detect human emotions.

• Journal of Broadcast Engineering
• /
• v.27 no.3
• /
• pp.273-282
• /
• 2022

MPEG-VCM(Video Coding for Machine) aims to standardize video codec for machines. VCM provides data sets and anchors, which provide reference data for comparison, for several machine vision tasks including object detection, object segmentation, and object tracking. The evaluation template can be used to compare compression and machine vision task performance between anchor data and various proposed video codecs. However, performance comparison is carried out separately for each machine vision task, and information related to performance evaluation of multiple machine vision tasks on a single bitstream is not provided currently. In this paper, we propose a performance evaluation method of a video codec for AI-based multi-tasks. Based on bits per pixel (BPP), which is the measure of a single bitstream size, and mean average precision(mAP), which is the accuracy measure of each task, we define three criteria for multi-task performance evaluation such as arithmetic average, weighted average, and harmonic average, and to calculate the multi-tasks performance results based on the mAP values. In addition, as the dynamic range of mAP may very different from task to task, performance results for multi-tasks are calculated and evaluated based on the normalized mAP in order to prevent a problem that would be happened because of the dynamic range.

• Korean Journal of Remote Sensing
• /
• v.38 no.5_2
• /
• pp.695-706
• /
• 2022

Medium and high-resolution optical satellites have proven their effectiveness in detecting wildfire areas. However, smoke plumes generated by wildfire scatter visible light incidents on the surface, thereby interrupting accurate monitoring of the area where wildfire occurs. Therefore, a technology to extract smoke in advance is required. Deep learning technology is expected to improve the accuracy of smoke extraction, but the lack of training datasets limits the application. However, for clouds, which have a similar property of scattering visible light, a large amount of training datasets has been accumulated. The purpose of this study is to develop a smoke extraction technique using deep learning, and the limits due to the lack of datasets were overcome by using a cloud dataset on transfer learning. To check the effectiveness of transfer learning, a small-scale smoke extraction training set was made, and the smoke extraction performance was compared before and after applying transfer learning using a public cloud dataset. As a result, not only the performance in the visible light wavelength band was enhanced but also in the near infrared (NIR) and short-wave infrared (SWIR). Through the results of this study, it is expected that the lack of datasets, which is a critical limit for using deep learning on smoke extraction, can be solved, and therefore, through the advancement of smoke extraction technology, it will be possible to present an advantage in monitoring wildfires.

• Korean Journal of Remote Sensing
• /
• v.37 no.2
• /
• pp.211-220
• /
• 2021

Land Surface Temperature (LST) is one of the useful parameters to diagnose the growth and development of crop and to detect crop stress. Unmanned Aerial Vehicle (UAV)-based LST (LSTUAV) can be estimated in the regional spatial scale due to miniaturization of thermal infrared camera and development of UAV. Given that meteorological variable, type of instrument, and surface condition can affect the LSTUAV, the evaluation for accuracy of LSTUAV is required. The purpose of this study is to evaluate the accuracy of LSTUAV using LST measured at ground (LSTGround) under various meteorological conditions and growth phases of garlic crop. To evaluate the accuracy of LSTUAV, Relative humidity (RH), absolute humidity (AH), gust, and vegetation index were considered. Root mean square error (RMSE) after minimizing the bias between LSTUAV and LSTGround was 2.565℃ under above 60% of RH, and it was higher than that of 1.82℃ under the below 60% of RH. Therefore, LSTUAV measurement should be conducted under the below 60% of RH. The error depending on the gust and surface conditions was not statistically significant (p-value < 0.05). LSTUAV had reliable accuracy under the wind speed conditions that allow flight and reflected the crop condition. These results help to comprehend the accuracy of LSTUAV and to utilize it in the agriculture field.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.25 no.4
• /
• pp.56-64
• /
• 2021

Leakage due to deterioration and damage is one of the major causes of volume change by freezing and thawing, and it leads micro-cracking and surface scaling in concrete structures. The deterioration of damaged concrete accelerates with the chloride attack. Thus, in the detailed guidelines for facility performance evaluation (2020), the quality of cover concrete and the freeze-thaw (FT) repetition cycle were newly suggested for concrete durability assessment. The quality of cover concrete should be evaluated by the rebound hammer test and the FT repetition cycle should be also considered in the deterioration environmental assessment. This study suggested the application of fast dynamic based nonlinear ultrasound method to monitor initial micro-scale damage under freezing and thawing environment. Concrete specimens were fabricated with different water-cement ratios (40%, 60%) and air contents (1.5% and 3.0%). The compressive strength, rebound number, relative dynamic modulus, and nonlinear ultrasound were measured with different FT cycles. The scanning electron microscopy was also performed to investigate the micro-scale FT damage. As a result, both the rebound number and the relative dynamic modulus had difficulty to detect early damage but the proposed method showed a potential to detect initial micro-scale damage and predict the FT resistance performance of concrete.

• Journal of Intelligence and Information Systems
• /
• v.29 no.3
• /
• pp.249-265
• /
• 2023

Recently, the satellite industry has been paying attention to the private-led 'New Space' paradigm, which is a departure from the traditional government-led industry. The space industry, which is considered to be the next food industry, is still receiving relatively little attention in Korea compared to the global market. Therefore, the purpose of this study is to explore future signals that can help determine the market entry strategies of private companies in the domestic satellite industry. To this end, this study utilizes the theoretical background of future signal theory and the Keyword Portfolio Map method to analyze keyword potential in patent document data based on keyword growth rate and keyword occurrence frequency. In addition, news data was collected to categorize future signals into first symptom and early information, respectively. This is utilized as an interpretive indicator of how the keywords reveal their actual potential outside of patent documents. This study describes the process of data collection and analysis to explore future signals and traces the evolution of each keyword in the collected documents from a weak signal to a strong signal by specifically visualizing how it can be used through the visualization of keyword maps. The process of this research can contribute to the methodological contribution and expansion of the scope of existing research on future signals, and the results can contribute to the establishment of new industry planning and research directions in the satellite industry.

• Geophysics and Geophysical Exploration
• /
• v.26 no.3
• /
• pp.126-137
• /
• 2023

In this study, a recurrent neural network (RNN) was employed as a methodological approach to classify dolphin click signals derived from ocean monitoring data. To improve the accuracy of click signal classification, the single time series data were transformed into fractional domains using fractional Fourier transform to expand its features. Transformed data were used as input for three RNN models: long short-term memory (LSTM), gated recurrent unit (GRU), and bidirectional LSTM (BiLSTM), which were compared to determine the optimal network for the classification of signals. Because the fractional Fourier transform displayed different characteristics depending on the chosen angle parameter, the optimal angle range for each RNN was first determined. To evaluate network performance, metrics such as accuracy, precision, recall, and F1-score were employed. Numerical experiments demonstrated that all three networks performed well, however, the BiLSTM network outperformed LSTM and GRU in terms of learning results. Furthermore, the BiLSTM network provided lower misclassification than the other networks and was deemed the most practically appliable to field data.

• Journal of Internet of Things and Convergence
• /
• v.10 no.2
• /
• pp.1-15
• /
• 2024

The construction of smart communities is a new method and important measure to ensure the security of residential areas. In order to solve the problem of low accuracy in face recognition caused by distorting facial features due to monitoring camera angles and other external factors, this paper proposes the following optimization strategies in designing a face recognition network: firstly, a global graph convolution module is designed to encode facial features as graph nodes, and a multi-scale feature enhancement residual module is designed to extract facial keypoint features in conjunction with the global graph convolution module. Secondly, after obtaining facial keypoints, they are constructed as a directed graph structure, and graph attention mechanisms are used to enhance the representation power of graph features. Finally, tensor computations are performed on the graph features of two faces, and the aggregated features are extracted and discriminated by a fully connected layer to determine whether the individuals' identities are the same. Through various experimental tests, the network designed in this paper achieves an AUC index of 85.65% for facial keypoint localization on the 300W public dataset and 88.92% on a self-built dataset. In terms of face recognition accuracy, the proposed network achieves an accuracy of 83.41% on the IBUG public dataset and 96.74% on a self-built dataset. Experimental results demonstrate that the network designed in this paper exhibits high detection and recognition accuracy for faces in surveillance videos.

• Journal of the korean academy of Pediatric Dentistry
• /
• v.27 no.1
• /
• pp.169-179
• /
• 2000

The purpose of this study is to develop the system which convert the optical difference of teeth texture between intact enamel and incipient caries lesion into shade difference by laser fluorescence and to develop new and simple caries activity test using laser fluorescence. The experimental design of this study consists of three parts. In first part, a new method for the in vitro assessment of changes in initial enamel caries lesion of Bovine teeth using laser fluorescence is tested. In second part, in vivo assessment undertaken. Number of teeth which showed incipient carious lesion on buccal surface examined by laser fluorescence was compared with the caries activity test of $Cariescreen^{(R)}$ test and other oral environmental test of dDfFtT. In third part, new caries activity test measured by laser fluorescence was developed on the basis of above results and evaluated the sensitivity, specificity, and diagnostic power. Optical density measured by laser fluorescence was increased as increasing the depth of incipient carious lesion and showed high correlation$(\gamma=0.7015)$ with lesion depth. Optical density showed direct proportion to lesion depth. Linear equation was obtained between the optical density and the lesion depth by regression analysis. The result of caries activity test with laser fluorescence showed high correlation with those of $Cariescreen^{(R)}$ test and dDfFtT examination. Caries activity test with laser fluorescence showed 48% of sensitivity, 52% of specificity, and 45% of diagnostic power on the basis of dDfFtT examination, and also showed 48% of sensitivity, 51% of specificity, and 36% of diagnostic power on the basis of $Cariescreen^{(R)}$ test. In regard above result, caries activity test with laser fluorescence considered to be reliable for caries activity test compared with other oral environmental test. and it was also considered to be practical because it would be simple, inexpensive, and time saving method.

• Sleep Medicine and Psychophysiology
• /
• v.10 no.1
• /
• pp.52-60
• /
• 2003

Objectives: Stage 1 sleep provides important information regarding interpretation of nocturnal polysomnography, particularly sleep onset. It is a short transition period from wakeful consciousness to sleep. The lack of prominent sleep events characterizing stage 1 sleep is a major obstacle in automatic sleep stage scoring. In this study, utilization of simultaneous EEG and EOG processing and analyses to detect stage 1 sleep automatically were attempted. Methods: Relative powers of the alpha waves and the theta waves were calculated from spectral estimation. A relative power of alpha waves less than 50% or relative power of theta waves more than 23% was regarded as stage 1 sleep. SEM(slow eye movement) was defined as the duration of both-eye movement ranging from 1.5 to 4 seconds, and was also regarded as stage 1 sleep. If one of these three criteria was met, the epoch was regarded as stage 1 sleep. Results were compared to the manual rating results done by two polysomnography experts. Results: A total of 169 epochs were analyzed. The agreement rate for stage 1 sleep between automatic detection and manual scoring was 79.3% and Cohen’s Kappa was 0.586 (p<0.01). A significant portion (32%) of automatically detected stage 1 sleep included SEM. Conclusion: Generally, digitally-scored sleep staging shows accuracy up to 70%. Considering potential difficulty in stage 1 sleep scoring, accuracy of 79.3% in this study seems to be strong enough. Simultaneous analysis of EOG differentiates this study from previous ones which mainly depended on EEG analysis. The issue of close relationship between SEM and stage 1 sleep raised by Kinnari remains a valid one in this study.