• ETRI Journal
• /
• v.38 no.5
• /
• pp.807-817
• /
• 2016

A novel fast algorithm is suggested for a coding unit (CU) mode decision using pseudo rate-distortion based on a separated encoding structure in High Efficiency Video Coding (HEVC). A conventional HEVC encoder requires a large computational time for a CU mode prediction because prediction and transformation procedures are applied to obtain a rate-distortion cost. Hence, for the practical application of HEVC encoding, it is necessary to significantly reduce the computational time of CU mode prediction. As described in this paper, under the proposed separated encoder structure, it is possible to decide the CU prediction mode without a full processing of the prediction and transformation to obtain a rate-distortion cost based on a suitable condition. Furthermore, to construct a suitable condition to improve the encoding speed, we employ a pseudo rate-distortion estimation based on a Hadamard transformation and a simple quantization. The experimental results show that the proposed method achieves a 38.68% reduction in the total encoding time with a similar coding performance to that of the HEVC reference model.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.2
• /
• pp.51-58
• /
• 2023

This paper proposes a deep learning method for estimating the heart rate from facial videos. Our proposed method estimates remote photoplethysmography (rPPG) signals to predict the heart rate. Although there have been proposed several methods for estimating rPPG signals, most previous methods can not be utilized in low-power single board computers due to their computational complexity. To address this problem, we construct a lightweight student model and employ a knowledge distillation technique to reduce the performance degradation of a deeper network model. The teacher model consists of 795k parameters, whereas the student model only contains 24k parameters, and therefore, the inference time was reduced with the factor of 10. By distilling the knowledge of the intermediate feature maps of the teacher model, we improved the accuracy of the student model for estimating the heart rate. Experiments were conducted on the UBFC-rPPG dataset to demonstrate the effectiveness of the proposed method. Moreover, we collected our own dataset to verify the accuracy and processing time of the proposed method on a real-world dataset. Experimental results on a NVIDIA Jetson Nano board demonstrate that our proposed method can infer the heart rate in real time with the mean absolute error of 2.5183 bpm.

• Clinical Endoscopy
• /
• v.55 no.3
• /
• pp.332-338
• /
• 2022

Continuous measurement of quality indicators (QIs) should be a routine part of colonoscopy, as a wide variation still exists in the performance and quality levels of colonoscopy in Korea. Among the many QIs of colonoscopy, the adenoma detection rate, average withdrawal time, bowel preparation adequacy, and cecal intubation rate should be monitored in daily clinical practice to improve the quality of the procedure. The adenoma detection rate is the best indicator of the quality of colonoscopy; however, it has many limitations for universal use in daily practice. With the development of natural language processing, the adenoma detection rate is expected to become more effective and useful. It is important that colonoscopists do not strictly and mechanically maintain an average withdrawal time of 6 minutes but instead perform careful colonoscopy to maximally expose the colonic mucosa with a withdrawal time of at least 6 minutes. To achieve adequate bowel preparation, documentation of bowel preparation with the Boston Bowel Preparation Scale (BBPS) should be a routine part of colonoscopy. When colonoscopists routinely followed the bowel preparation protocols, ≥85% of outpatient screening colonoscopies had a BBPS score of ≥6. In addition, the cecal intubation rate should be ≥95% of all screening colonoscopies. The first step in improving colonoscopy quality in Korea is to apply these key performance measurements in clinical practice.

• International Journal of Internet, Broadcasting and Communication
• /
• v.12 no.3
• /
• pp.125-130
• /
• 2020

At the end of 1997, the volatility of the exchange rate intensified as the nation's exchange rate system was converted into a free-floating exchange rate system. As a result, managing the exchange rate is becoming a very important task, and the need for forecasting the exchange rate is growing. The exchange rate prediction model using the existing exchange rate prediction method, statistical technique, cannot find a nonlinear pattern of the time series variable, and it is difficult to analyze the time series with the variability cluster phenomenon. And as the number of variables to be analyzed increases, the number of parameters to be estimated increases, and it is not easy to interpret the meaning of the estimated coefficients. Accordingly, the exchange rate prediction model using artificial neural network, rather than statistical technique, is presented. Using DNN, which is the basis of deep learning among artificial neural networks, and LSTM, a recurrent neural network model, the number of hidden layers, neurons, and activation function changes of each model found the optimal exchange rate prediction model. The study found that although there were model differences, LSTM models performed better than DNN models and performed best when the activation function was Tanh.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.53 no.6
• /
• pp.85-91
• /
• 2011

Three different types of wetlands (unplanted wetland, reed planted wetland, cattail planted wetland) were constructed at the mouth of Seokmoon reservoir with 910 $m^2$ each to examine the effects of wetland plant on pollutant removal rate in constructed wetland, and operated for 9 years (2002~2010). Water depth of the wetland was maintained at 0.3~0.5 m, flow rate was about 40~200 $m^3$/day, and retention time was managed at about 1~5 days. There was no difference in removal rate of SS, TN, and TP between reed wetland and cattail wetland. Removal rate of SS and TN in planted wetland with reed and cattail were higher than unplanted wetland, whereas removal rate of TP in unplanted wetland was higher then planted wetland. The monthly variation of removal rate in planted wetlands was high compared with unplanted wetland. From the long term monitoring results, SS and TN removal rates of period3 (2008~2010) were higher than period1 (2002~2004) in planted wetland, whereas TP removal rate was decreased as time goes on. Overall, pollutant removal rate in constructed wetland was more influenced by existence of plants than by plant species. Although constructed wetland is operated long term period, SS, TN, and TP removal rate (SS 90 %, TN 60 %, TP 40 %) can be maintained high values.

• Journal of Embryo Transfer
• /
• v.13 no.1
• /
• pp.69-76
• /
• 1998

This study was carried out to establish an effective system for embryo transfer techniques by analyzing several factors affecting in-vivo embryo transfer in Korean cattle Embryos were transferred into a toral of 301 recipients. The results obtained in studies on the factors affecting pregnancy rate after embryo transfer by condition of transfer time were as follows ; 1. The pregnancy rate by the seasons of transferred fresh and frozen embryos were not different, but the pregnancy rate was slightly higher in summer(80.8%). 2. The pregnancy rate by the days of embryo transfer after estrus were not different when fresh embryos were transferred, but the pregnancy rate was highest at 8 days when frozen embryos were transferred(P<0.01, 40.0%). 3. The pregnancy rate at estrus synchronization was remarkably higher with PGF$_2$$\alpha$ treated than natural (P<0.05, 70.4%, 43.4%). 4. The pragnancy rate by the degree of estrus synchronization was best when the estrus was synchronized in both fresh and frozen embryos (83.3% and 29.7%, respectively), but the pregnancy rate was not different among $\pm$2 days. But the pregnancy rate of frozen embryos were slightly higher when the recipients exhibited estrus earlier than donors.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.46 no.3
• /
• pp.209-214
• /
• 2001

Physiological responses of rice to the flooding time of different water turbidity (clear water, sub-muddy water, muddy water) were analyzed as photosynthesis, chlorophyll fluorescence, transpiration, and physiological recovery. Photosynthetic rate was higher as turbidity increased and decreased as flooding time was extended. Floodings of 36 hrs and 42 hrs were resulted in 25% and 50% decrease of photosynthetic rate, respectively. Transpiration rate was higher in high turbidity (increased 30%, 25%, and 20% in clear, sub-muddy, and muddy water, respectively) and in increased floodings. Fv/Fm decreased as increased turbidity and flooding time. About 20% decrease of Fv/Fm was recorded in 48 hrs and 36 hrs after flooding with clear water and sub-muddy water(including muddy water), respectively. Total nitrogen was decreased with flooding treatment. Significant decrease of total nitrogen was occurred 36 hrs after flooding with muddy water, Dry weight measured 2 weeks after flooding treatment as an indication of recovery of flooding stress didn't show significant difference with turbidity, but significantly decreased as flooding time was prolonged. About 25% and 50% decreases were found in 24 hrs and 42 hrs flooding time, respectively. furthermore, 48 hrs of flooding with sub-muddy and muddy water resulted in no physiological recovery. Photosynthetic rate was decreased 15% and 10% with clear water and muddy water(including sub-muddy water), respectively. The rate was dramatically decreased 42 hrs after flooding. Transpiration rate increased about 20% regardless of turbidity and flooding time. We found transitory decrease of photosynthetic and transpiration rate at the point of 24 hrs after flooding and right after do-flooding.

• The Korean Journal of Physiology
• /
• v.5 no.2
• /
• pp.15-27
• /
• 1971

Physiological analysis of the physical exercise was made on 9 subjects performing mountain climbing. The course between two points (256 and 516 meters altitude) was 1,300 meters in distance and difference of vertical height was 260 meters making the mean grade of 20%. In the field, the heart rates during uphill or downhill walk were recorded by EKG radio-telemetry. In the laboratory, oxygen consumption was obtained by the recorded heart rates, using individual heart rate vs oxygen consumption diagram obtained by treadmill test. the following results were obtained. 1. Uphill walk time was 36.5 minutes, and during this period the mean heart rate was 149.0 heats/min and peak heart rate was 169.2 beats/min. The total heart beats during the uphill walk was 5.433 beats. 2. The ratio of individual mean heart rate during the uphill walk to the maximal heart rate distributed between 66.6% and 98.3%, and the mean of the total group was 83.1%. The ratio of peak heart rate of uphill walk to the maximal heart rate was 94.5% in the group. Thus uphill walk of a 20% grade mountain course was an exhaustive exercise. 3. Oxygen consumption during uphill walk was 2.22 l/min (ranged between 1.79 and 2.70 l/min) and the ratio of this to the resting oxygen consumption was 8.31. The peak value of oxygen consumption during uphill walk was 2.73 l/min and the ratio of this to the resting oxygen consumption was 10.39. 4. Energy expenditure during uphill walk showed a mean of 11.1 kcal/min and the peak expenditure rate was 13.6 kcal/min. The total energy expenditure during 36.5 minutes of uphill walk was 396 kcal. 5. In downhill walk, the time was 31.7 minutes, mean heart rate was 118.4 (ranged between 100.1 and 142.7) beats/min, and the peak heart rate was only 129.4 beats/min. The ratio of mean heart rate to the maximal heart rate was 66.3%. Total heart beats during downhill walk was 3,710 beats. The ratio of downhill oxygen consumption to the resting consumption was 5.70. The rate of energy expenditure was 7.5 kcal/min, and the total onery expenditure during the 31.7 minutes of downhill walk was 228 kcal. 6. The effect of training was manifest in the uphill walk and not in the downhill walk. After training in mountain course walk, i) the uphill time was shortened, ii) mean heart rate increased, iii) time vs heart rate curve became smooth and showed less frequent zig-zag, i.e., the depth of trough on the curve decreased and the magnitude was less than 10 beats. In non-trained subject the depth of trough on the curve was greater than 50 beats and appeared more frequently. 7. Mountain climbing is a good health promotion exercise. For the promotion of health the reasonable amount of uphill mountain walk exercise in a 20% grade course is a walk for 40 or 50 minutes duration once a week.

• Journal of Korean Society for Quality Management
• /
• v.17 no.2
• /
• pp.17-26
• /
• 1989

This paper is concerned with forecasting the existing number of errors in the computer software and optimizing the stopping time of the software test based upon the forecasted number of errors. The most commonly used models have assessed software reliability under the assumption that the software failure late is proportional to the current fault content of the software but invariant to time since software faults are independents of others and equally likely to cause a failure during testing. In practice, it has been observed that in many situations, the failure rate decrease. Hence, this paper proposes a mathematical model to describe testing situations where the failure rate of software limearly decreases proportional to testing time. The least square method is used to estimate parameters of the mathematical model. A cost model to optimize the software testing time is also proposed. In this cost mode two cost factors are considered. The first cost is to test execution cost directly proportional to test time and the second cost is the failure cost incurred after delivery of the software to user. The failure cost is assumed to be proportional to the number of errors remained in the software at the test stopping time. The optimal stopping time is determined to minimize the total cost, which is the sum of test execution cast and the failure cost. A numerical example is solved to illustrate the proposed procedure.

• Proceedings of the KSRS Conference
• /
• v.2
• /
• pp.635-638
• /
• 2006

Regression of conventional prediction techniques in data mining uses the model which is generated from the training step. This model is applied to new input data without any change. If this model is applied directly to time series, the rate of prediction accuracy will be decreased. This paper proposes an incremental regression for time series prediction like typhoon track prediction. This technique considers the characteristic of time series which may be changed over time. It is composed of two steps. The first step executes a fractional process for applying input data to the regression model. The second step updates the model by using its information as new data. Additionally, the model is maintained by only recent data in a queue. This approach has the following two advantages. It maintains the minimum information of the model by using a matrix, so space complexity is reduced. Moreover, it prevents the increment of error rate by updating the model over time. Accuracy rate of the proposed method is measured by RME(Relative Mean Error) and RMSE(Root Mean Square Error). The results of typhoon track prediction experiment are performed by the proposed technique IMLR(Incremental Multiple Linear Regression) is more efficient than those of MLR(Multiple Linear Regression) and SVR(Support Vector Regression).