• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.7
• /
• pp.2407-2424
• /
• 2022

Recently, Transformer has made great progress in video retrieval tasks due to its high representation capability. For the structure of a Transformer, the cascaded self-attention modules are capable of capturing long-distance feature dependencies. However, the local feature details are likely to have deteriorated. In addition, increasing the depth of the structure is likely to produce learning bias in the learned features. In this paper, an improved Transformer structure named TransDCS (Transformer with Dynamic Convolution and Shortcut) is proposed. A Multi-head Conv-Self-Attention module is introduced to model the local dependencies and improve the efficiency of local features extraction. Meanwhile, the augmented shortcuts module based on a dual identity matrix is applied to enhance the conduction of input features, and mitigate the learning bias. The proposed model is tested on MSRVTT, LSMDC and Activity-Net benchmarks, and it surpasses all previous solutions for the video-text retrieval task. For example, on the LSMDC benchmark, a gain of about 2.3% MdR and 6.1% MnR is obtained over recently proposed multimodal-based methods.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.5 no.3 s.11
• /
• pp.1-12
• /
• 2006

Nowdays ITS is being installed in each local autonomous entity, and BIS installation and operation is prior to other ITS sub systems for the public service. The methods of positioning md wireless communication in BIS are DSRC+DSRC, GPS+wireless communication, Beacon+Beacon, which are chosen and operated as the local features. Before this study, the before and after survey of BIS' quality have only been done without performance evaluation of BIS. So the method of BIS' evaluation have been established including performance test in this paper. And the evaluation of some BIS' reliability and efficiency have been done using the reliability of arriving data and the wireless communication response rates after choosing typical BIS' sub system.

• Journal of Information Display
• /
• v.11 no.4
• /
• pp.173-181
• /
• 2010

In this paper, light-emitting-diode (LED) backlight driving circuits and dimming method for medium-sized and large liquid crystal displays (LCDs) are proposed. The double loop control method, the intelligent-phase-shifted PWM dimming method, the fast-switching current regulator, and the current matching techniques are proposed to improve not only the current regulation characteristics and the power efficiency but also the current matching characteristics and the transient response of the LED current. The brightness of the backlight using the proposed local dimming method was determined from the histogram of the local block to reduce the power consumption of the backlight without image distortion. The measured maximum power efficiency of the LED backlight driving circuit for medium-sized LCDs was 90%, and the simulation results showed an 88% maximum power efficiency of the LED backlight driving circuit for large LCDs. The maximum backlight power-saving ratio of the proposed dimming method was 41.7% in the simulation with a high-contrast image. The experiment and simulation results showed that the performance of LEDs as LCD backlight units (BLUs) improved with the proposed circuits and method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.11C
• /
• pp.1066-1070
• /
• 2006

In this paper, we present a residual filter to improve performance of H.264 compressed video. In general, noisy video sequences captured by imaging system result in decline of coding efficiency and unpleasant coding artifacts due to higher frequency components. By incorporating local statistics and quantization parameter into filtering process, the spurious noise is significantly attenuated and coding efficiency is improved, leading to improvement of visual quality and to bit-rate saving for given quantization step size. In addition, in order to reduce the complexity of the residual filter, noise induced by analyzing H.264 transformation and quantization processes are introduced. The simulation results show the capability of the proposed algorithm.

• Journal of Korean Society for Quality Management
• /
• v.37 no.1
• /
• pp.1-9
• /
• 2009

Data envelopment analysis(DEA) has proved to be powerful for benchmarking and has been widely used in a variety of settings since the advent of it. DEA can be used in identifying the best performing units to be benchmarked against as well as in providing actionable measure for improvement of a organization's performance. However, the selection of performance benchmarks is a matter of both technical production possibilities and organizational policy considerations, managerial preferences and external restrictions. In that regards, DEA has a limited value in benchmarking because it focuses on only technical production Possibilities. This research proposes a new perspective in using DEA and a frame-work for benchmarking to select benchmarks that are both feasible and desirable in terms of organizational context. To do this, the concept of local and global efficiency is newly proposed. To show how useful the suggested concept and framework are, a case study is addressed.

• 한국연소학회:학술대회논문집
• /
• 2006.04a
• /
• pp.19-25
• /
• 2006

Energy efficient and low pollution combustion systems the use gaseous fuels have been in great demand in recent year. Radiant burner in many different forms are emerging as very desirable combustion systems for same reason. Porous radiant burners are used in drying, preheating and curing, and in other type of materials processing and manufacturing processes. However, little knowledge is available about the operating characteristics and the structure of flames in porous ceramic fiber radiant burners. The objective of the present work is to investigate the global performance characteristics of the ceramic fiber burner. A detailed study which includes the spectral intensity, gas temperature, radiation efficiency and global pollutant emissions. Another objective is to study the flame structure of the ceramic fiber burner by measuring the local gas temperature. The results indicate that ceramic fiber burner do offer a 19-44% gain in radiant efficiency. The ceramic fiber burner exhibit significant spectral intensity peaks in the band at $2.0-2.5{\mu}m$. The local temperature distribution inside the mat and near the mat surface as a function of the equivalence ratio can be reasonably interpreted by the relation of the heat balance in the mat and movement of the reaction zone. Nox emission from ceramic fiber burner is less than 25ppm throughout the operating range.

• Laser Solutions
• /
• v.14 no.3
• /
• pp.1-6
• /
• 2011

Globally, the interest of renewable energy has become an upsurge. Especially, the solar industry is the one which is getting rapid growth rate. Many of researchers have been undertaking to improve the efficiency of solar cell to accomplish grid parity. The most of research has been concentrated on two methods, one on the selective emitter and the other is on LBSF (Local Back Surface Field) formation. Laser patterning will be needed to eliminate the thin film to form selective emitter and LBSF of solar cell. This paper reports some experimental results in laser patterning process for high-efficiency crystalline solar cell manufacturing. The experimental results indicate that the patterning quality depends on the average power and repetition rate of laser. The experimental results prove that the laser patterning process is an advantageous method to improve the efficiency of solar cell.

• Journal of computational fluids engineering
• /
• v.11 no.4 s.35
• /
• pp.1-8
• /
• 2006

Clocking effects of a stator on the performance and internal flow in an UTRC 1.5 stage axial turbine are investigated using a three-dimensional unsteady flow simulation. Six relative positions of two rows of stator are investigated by positioning the second stator being clocked in a step of 1/6 pitch. The relative efficiency benefit of about 1% is obtained depending on the clocking positions. However, internal flows have some different characteristics from that in the previous study at the best and worst efficiency positions, since the first stator wake is mixed out with the rotor wake before arriving at the leading edge of the second stator. Instead of the first stator wake, it is found that the wake interaction of the first stator and rotor has a important role on a relative efficiency variation at each clocking position. The time-averaged local efficiency along the span at the maximum efficiency is more uniform than that at the minimum efficiency. That is, the spanwise efficiency distribution at the minimum efficiency has larger values in mid-span but smaller values near the hub and casing in comparison to those at the maximum efficiency. Moreover, the difference between maximum and minimum instantaneous efficiencies during one period is found to be smaller at the maximum efficiency than at the minimum efficiency.

• 유체기계공업학회:학술대회논문집
• /
• 2005.12a
• /
• pp.473-480
• /
• 2005

Clocking effects of a stator on the performance and internal flow in an UTRC 1.5 stage axial turbine are investigated using a three-dimensional unsteady flow simulation. Six relative positions of two rows of stator are investigated by positioning the second stator being clocked in a step of 1/6 pitch. The relative efficiency benefit of about 1% is obtained depending on the clocking positions. However, internal flows have some different characteristics from that in the previous study at the best and worst efficiency positions, since be first stator wake is mixed out with the rotor wake before arriving at the leading edge of the second stator. Instead of the first stator wake, it is found that the wake interaction of the first stator and rotor has a important role on a relative efficiency variation at each clocking position. The time-averaged local efficiency along the span at the maximum efficiency is more uniform than that at the minimum efficiency. That is, the spanwise efficiency distribution at the minimum efficiency has larger values in mid-span but smaller values near the hub and casing in comparison to those at the maximum efficiency. Moreover, the difference between maximum and minimum instantaneous efficiencies during one period is found to be smaller at the maximum efficiency than at the minimum efficiency.

• Nuclear Engineering and Technology
• /
• v.35 no.3
• /
• pp.234-242
• /
• 2003

The full energy peak efficiency of a hyper pure germanium (HPGe) detector was calibrated in a wide energy range from 0.06 to 11 MeV. Both the experimental technique and the Monte Carlo method were used for the efficiency calibration. The measurement was performed using the standard radioisotopes in the low energy region of $60{\sim}1408$ keV, which was further extended up to 11 MeV by using the $^{14}N(n,r)\;and\;^{35}Cl(n,r)$ reactions. The GEANT Monte Carlo code was used for efficiency calculation. The calculated efficiency had the same dependency on the r-ray energy with the measurement, and the discrepancy between the calculation and the measurement was minimized by fine-tuning of the detector geometry. From the calculated result, the efficiency curve of the HPGe detector was reliably determined particularly in the high energy region above several MeV, where the number of measured efficiency points is relatively small despite the wide energy region. The calculated efficiency agreed with the measurement within about $7\%$. In addition to the efficiency calculation, the origin of the local minimum near 600 keV on the efficiency curve was analyzed as a general characteristics of a HPGe detector.