• Journal of Korea Multimedia Society
• /
• v.24 no.7
• /
• pp.890-902
• /
• 2021

Although the object detection accuracy with still images has been significantly improved with the advance of deep learning techniques, the object detection problem with video data remains as a challenging problem due to the real-time requirement and accuracy drop with occlusion. In this research, we propose a method in pig detection for video monitoring environment. First, we determine a motion, from a video data obtained from a tilted-down-view camera, based on the average size of each pig at each location with the training data, and extract key frames based on the motion information. For each key frame, we then apply YOLO, which is known to have a superior trade-off between accuracy and execution speed among many deep learning-based object detectors, in order to get pig's bounding boxes. Finally, we merge the bounding boxes between consecutive key frames in order to reduce false positive and negative cases. Based on the experiment results with a video data set obtained from a pig farm, we confirmed that the pigs could be detected with an accuracy of 97% at a processing speed of 37fps.

• Nuclear Engineering and Technology
• /
• v.41 no.7
• /
• pp.885-892
• /
• 2009

This manuscript will discuss a numerical method where the six equations of two-phase flow, the solid heat conduction equations, and the two equations that describe neutron diffusion and precursor concentration are solved together in a tightly coupled, nonlinear fashion for a simplified model of a nuclear reactor core. This approach has two important advantages. The first advantage is a higher level of accuracy. Because the equations are solved together in a single nonlinear system, the solution is more accurate than the traditional "operator split" approach where the two-phase flow equations are solved first, the heat conduction is solved second and the neutron diffusion is solved third, limiting the temporal accuracy to $1^{st}$ order because the nonlinear coupling between the physics is handled explicitly. The second advantage of the method described in this manuscript is that the time step control in the fully implicit system can be based on the timescale of the solution rather than a stability-based time step restriction like the material Courant limit required of operator-split methods. In this work, a pilot code was used which employs this tightly coupled, fully implicit method to simulate a reactor core. Results are presented from a simulated control rod movement which show $2^{nd}$ order accuracy in time. Also described in this paper is a simulated rod ejection demonstrating how the fastest timescale of the problem can change between the state variables of neutronics, conduction and two-phase flow during the course of a transient.

• Journal of Hydrogen and New Energy
• /
• v.33 no.6
• /
• pp.692-698
• /
• 2022

With the expansion of domestic hydrogen fuelling station infrastructure, it is necessary to secure reliability among hydrogen traders, and for this, technology to accurately measure hydrogen is important. In this study, 4 types of hydrogen trading volume calculation models (model 1-4) were presented to improve the accuracy of the hydrogen trading volume. In order to obtain the reference value of model 4, and experiment was conducted using a flow rate measurement equipment, and the error rate of the calculated value for each model was compared and analyzed. As a result, model 1 had the lowest metering accuracy, model 2 had the second highest metering accuracy and model 3 had the highest metering accuracy until a certain point. But after the point, model 2 had the highest metering accuracy and model 3 had the second metering accuracy.

• English Language & Literature Teaching
• /
• v.13 no.2
• /
• pp.63-90
• /
• 2007

The present study investigated what combination of three form-focused options - explicit explanation, production practice, and corrective feedback - may be effective in helping low-proficiency learners improve accuracy in communicative writing. The subjects were 34 Korean university students enrolled in 'Business English 1' and the study lasted 11 weeks. The relative clause structure was selected as the target structure. The study found that the combination of explicit explanation, sentence-level production practice, communicative writing practice, and recasts had a significantly greater effect on improved accuracy than the combination of communicative writing practice and recasts and that of explicit explanation, communicative writing practice, and recasts. Because the second and third combinations didn't lead to significantly improved accuracy, it can be concluded that of the form-focused options forming the first combination sentence-level production practice made a decisive contribution to the significant increase in accuracy. It also found that the provision of self-correcting opportunities before providing recasts on errors committed in sentence-level production practice resulted in significantly greater accuracy in communicative writing than the provision of recasts alone on them. The results of the study suggest that we should make low-proficiency Korean learners have sentence-level production practice which is intensive and focused and make them self-correct targeted errors before providing them with narrowly focused recasts in order to help them to improve writing accuracy.

• Journal of Institute of Control, Robotics and Systems
• /
• v.20 no.2
• /
• pp.175-179
• /
• 2014

The histogram specification turns the shape of a histogram into that we want to specify. This technique can be applied usefully in various image processing fields such as machine vision. However, the histogram specification technique has its basic limits. For instance, the histogram does not have location information of pixels. Also, the accuracy of the specification drops because of quantization errors of the digitized image. In this paper, we proposed a multiresolution histogram specification method in order to improve the accuracy of specification in terms of resemblance between destination and source image. The experimental results show that the proposed method enhances the accuracy of the specification compared to the conventional methods.

• International journal of advanced smart convergence
• /
• v.13 no.2
• /
• pp.80-87
• /
• 2024

This paper focuses on improving accuracy in constrained computing settings by employing the ReLU (Rectified Linear Unit) activation function. The research conducted involves modifying parameters of the ReLU function and comparing performance in terms of accuracy and computational time. This paper specifically focuses on optimizing ReLU in the context of a Multilayer Perceptron (MLP) by determining the ideal values for features such as the dimensions of the linear layers and the learning rate (Ir). In order to optimize performance, the paper experiments with adjusting parameters like the size dimensions of linear layers and Ir values to induce the best performance outcomes. The experimental results show that using ReLU alone yielded the highest accuracy of 96.7% when the dimension sizes were 30 - 10 and the Ir value was 1. When combining ReLU with the Adam optimizer, the optimal model configuration had dimension sizes of 60 - 40 - 10, and an Ir value of 0.001, which resulted in the highest accuracy of 97.07%.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2003.06a
• /
• pp.1243-1248
• /
• 2003

It is very important to measure linear motion accuracy of NC lathe as it affects all other parts of machines machined by them in industries. If the motion accuracy of NC lathe is bad, the dimension accuracy and the change-ability of works will be bad in the assembly of machine parts. In this paper, computer software systems are organized to measure linear motion of ATC(Automatic tool changer) on zx plane of NC lathe using two linear scales and the time pulses coming out from computer in order to get data at constant time intervals from the linear scales. And each sets of error data obtained from the test is discripted to plots and the results of linear motion errors are expressed as numerics by computer treatment.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.15 no.6
• /
• pp.38-43
• /
• 2006

A slot cut in end milling processes is one of the laborious works because the cutting force is likely to deflect the tools excessively, then to make large errors or to fracture the tool. This difficulty is owing to the poor stiffness of slender shaped end mills. Though, in most cases, additional finish cuts are followed after rough cuts, the accuracy of rough cuts is still important because it affects the final accuracy after finish cuts and productivity. The accuracy in slot cuts depends on the tool stiffness and the cutting conditions including depth of cut and feed. In order to meet the desired accuracy, diameter of end mill and cutting allowance have to be selected carefully. This study suggests several guidances for selecting the end mill diameter and the slot cut allowance to improve machining accuracy and productivity in slot end millings. Some experiments were done with the various cutting parameters of tool diameter, depth of cut and feed.

• Design & Manufacturing
• /
• v.2 no.1
• /
• pp.11-14
• /
• 2008

we can save a development cost and time as computer was used in tool and die design of car fields in die manufacture process. Dimension accuracy errors such as springback, springgo, overcrown and twist were reduced product accuracy and caused trouble to assembly each parts of car. In this paper, CADCEUS was used to modify and optimize results of deflection for a tail gate panel of car parts in order to reduce dimension accuracy errors by springback in sheet metal forming. As CADCEUS was used to apply for a tail gate panel, the time for quality to improve was reduced to 30%.

• Nuclear Engineering and Technology
• /
• v.52 no.10
• /
• pp.2313-2319
• /
• 2020

The goal of this project is to achieve an accurate segmentation of the pulmonary tumors besides shortening the time and increasing the accuracy. Here, improved region growing (IRG) algorithm is introduced in order to segment the lung tumor with a sufficient accuracy in a shorter time compared to the other basics methods. This comprehensive algorithm was applied on 4 patients CT images and the results of the various steps on segmentation improvement shown 98% accuracy as compared to the basic algorithm. The combination of "multipoint growth start" produced a desirable outcome in accurately bounding the tumor. The proposed algorithm improved tumor identification by less than 13% along with a sufficient percentage of compliance accuracy.