• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.05a
• /
• pp.429-432
• /
• 2021

스마트 기기와 애플리케이션의 테스트를 위해 빠르고 정확하게 스마트 기기 화면 상에서 테스트가 필요한 위치를 추출해야 한다. 필요한 위치를 추출할 때 스마트 기기 화면과 테스트할 수 있는 영역의 매칭 방식을 사용하는데 이를 위해 이미지의 변형이 발생해도 원하는 영역의 matching point 을 빠르고 정확하게 추출하는 feature matching 방식의 D2-Net 의 feature extraction 모델과 fitting algorithm 을 변경하였다.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.38 no.2
• /
• pp.172-178
• /
• 2002

This study presents a dynamic simulation of a purse seine net behavior Mathematical model suitable for purse seining, which is based on data from a series of previous simulations, various field experiments, is modelized as a set of mass-spring system. In this model, a number of meshes are approximated as one mass point, each of which connected to its neighbors by massless springs, the equations of motion are derived from considering internal force from the springs and external forces such as resistance and gravitation. This simulation shows the quantitative state on every mass point of the net and purse line during the shooting and pursing phases. So it is possible that performance of a purse seine net be analyzed using various and evolving parameters such as the shooting speed, the hauling speed, the size or type of the sinker, float and twine, also the hanging ratio etc.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.1
• /
• pp.1-8
• /
• 2018

Recently, the convolutional neural network shows high performance in many computer vision tasks. However, convolutional neural networks require enormous amount of operation, so it is difficult to adopt them in the embedded environments. To solve this problem, many studies are performed on the ASIC or FPGA implementation, where an efficient representation method is required. The fixed-point representation is adequate for the ASIC or FPGA implementation but causes a performance degradation. This paper proposes a separate optimization of representations for the convolutional layers and the batch normalization layers. With the proposed method, the required bit width for the convolutional layers is reduced from 16 bits to 10 bits for the ResNet-50 neural network. Since the computation amount of the convolutional layers occupies the most of the entire computation, the bit width reduction in the convolutional layers enables the efficient implementation of the convolutional neural networks.

• International Journal of Contents
• /
• v.10 no.3
• /
• pp.84-89
• /
• 2014

The mass-spring model has been typically employed in physical-based simulators for clothes or patches. The mass-spring model frequently utilizes equal mass and the gravity factor. The model structure of masses supports a shape applicable to fishing nets. Therefore, to create a simulation model of a fishing net, we consider the mass-spring model and adopt the tidal-current and buoyancy effects in underwater environments. These additional factors lead to a more realistic visualization of fishing-net simulations. In this paper, we propose a new mass-spring model for a fishing-net and a method to simplify the calculation equations for a real-time simulation of a fishing-net model. Our 3D mass-spring model presents a mesh-structure similar to a typical mass-spring model except that each intersection point can have different masses. The motion of each mass is calculated periodically considering additional dynamics. To reduce the calculation time, we attempt to simplify the mathematical equations that include the effect of the tidal-current and buoyancy. Through this research, we expect to achieve a real-time and realistic simulation for the fishing net.

• KIISE Transactions on Computing Practices
• /
• v.20 no.11
• /
• pp.610-614
• /
• 2014

Like many types of scientific data, results from simulations of volcanic ash diffusion are of a clustered sparse matrix in the netCDF format. Since these data sets are large in size, they generate high storage and transmission costs. In this paper, we suggest a new method that reduces the size of the data of volcanic ash diffusion simulations by converting the multi-dimensional index to a single dimension and keeping only the starting point and length of the consecutive zeros. This method presents performance that is almost as good as that of ZIP format compression, but does not destroy the netCDF structure. The suggested method is expected to allow for storage space to be efficiently used by reducing both the data size and the network transmission time.

• Korean Journal of Plant Resources
• /
• v.8 no.2
• /
• pp.195-199
• /
• 1995

This study was conducted to investigate the Net photosynthesis and respiration rates among the varieties of Soybean(Eunha, Pangsa and Paldal that have high yields) at various leaf temperature and light intensity at the stage of $V_5$. The relations between the Net photosynthesis rate and SLW(specific leaf weight) and chlorophyll content were also investigated. 1. Net photosynthesis rates at $25^{\circ}C$ were $21.5mgdm^{-2}h^{-1}$ in cv. Eunha, $20.2mgdm^{-2}h^{-1}$ in cv. Pangsa and $18.5mgdm^{-2}h^{-1}$ in cv. Paldal. 2. Most cultivars of Soybean showed the maximum rates of Net photosynthesis at $25^{\circ}C$, especially in cv. Eunha. Also Net photosynthesis rates differed depending on the leaf shape. Long leaf shape(cv. Eunha) was better than round leaf shape(cv. Paldal) in Net photosynthesis rate. 3. Respiration rates of leaves in Eunha, Pangse and Paldal were $0.56mgdm^{-2}h^{-1}$ at $15^{\circ}C$, $0.79mgdm^{-2}h^{-1}$ at $20^{\circ}C$ $1.15mgdm^{-2}h^{-1}$ at $25^{\circ}C$ and $1.37mgdm^{-2}h^{-1}$ at $30^{\circ}C$. 4. Specific leaf weight were $3.1mg/cm^2$ in Pangsa, $3.5mg/cm^2$ in Eunha and Paldal. No signlficant difference were showed in net photosynthesis rates and specific lear weight. 5. Leaf chlorophyll content were $2.48{\mu}g/gF.W.$ in Eunha, $2.19{\mu}g/gF.W.$ in Pangsa and $1.67{\mu}/g F.W.$ in Paldal. Significant difference were showed in Net photosynthesis rates and Leaf chlorophyll content. 6. The estimated compensation points at which net photosynthesis approached zero were $10{\mu}Em^{-2}s^{-1}$ in Eunha, Pangsa, and Palda at 1$5^{\circ}C$. The compensation point in cv. Eunha at $20^{\circ}C$ was $12P{\mu}Em^{-2}s{-1}$ while $13{\mu}Em^{-2}s{-1}$ in Pangsa and Palda. The compensation point in cv. Paldal at $25^{\circ}C$ was $18{\mu}Em^{-2}s{-1}$ while $16{\mu}Em^{-2}s{-1}$ in Eunha and Pangsa. The compensation point in cv. Palda at $30^{\circ}C$ was $23{\mu}Em^{-2}s{-1}$ Palda while $13{\mu}Em^{-2}s{-1}$ in Eunha and Pangsa.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.26 no.10
• /
• pp.1961-1970
• /
• 2002

A laser scanner is widely used fur a device fur acquiring point data in reverse engineering. It is more efficient to generate a surface automatically from the line-typed data than scattered data of points clouds. In the case of a compound model, it is hard to represent all the scanned data into one surface maintaining its original line characteristics. In this paper, a method is presented to generate a surface by the segmentation of measured point data. After forming triangular net, the segmentation is done by the user input such as the angle between triangles, the number of facets to be considered as small segment, and the angle for combining small segment. B-spline fitting is implemented to the point data in each segment. The surface generation through segmentation shows a reliable result when it is applied to the models with curvature deviation regions. An useful algorithm for surface reconstruction is developed and verified by applying an practical model and shows a good tools fur reverse engineering in design modification.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.15 no.2
• /
• pp.83-94
• /
• 1979

It is more than half a century since anchovy boat seine has been introduced in Korean fishery to catch anchovies, but the study on it was began in 1970's by the authors. In 1971, the authors carried out an experiment about the net formation of the traditional gear in tow by using model net, and in 1978, about the patti net gear, commercially used in Japan. Now, the authors investigated the new model net, model 79, expecting to be suitable for commercial fishery in Korea, with the strong point of those two gears kept and the weak point of them corrected. The experimental gear was constructed attached the long net pendants to the fore end of extension wing by shortening its length in two-third of the traditional gear. Inside wing was improved so as to show high opening in tow. Rubber bobbins and hanging rings are used to prevent the heavy friction of bosom ground rope against the sea bed. The gear was used to catch anchovies in the commercial fishing ground in the south-eastern coastal waters of Korea, from May to October in 1979. From the experiment, the following results are found. 1. In opening height, the experimented gear was 30 percent greater than the traditional one. 2. It took 3 to 5 minutes for the bosom ground rope to sink from the surface to the sea bed, while 10 to 15 minutes for the traditional gear to do. 3. Ground rope never scooped mud, even in the muddy sea bed. 4. The gear showed better catchability than the traditional gear.

• Journal of Biomedical Engineering Research
• /
• v.44 no.2
• /
• pp.139-146
• /
• 2023

Osteoporosis is a disease in which the risk of bone fractures increases due to a decrease in bone density caused by aging. Osteoporosis is diagnosed by measuring bone density in the total hip, femoral neck, and lumbar spine. To accurately measure bone density in the lumbar spine, the vertebral region must be segmented from the lumbar X-ray image. Deep learning-based automatic spinal segmentation methods can provide fast and precise information about the vertebral region. In this study, we used 695 lumbar spine images as training and test datasets for a deep learning segmentation model. We proposed a lumbar automatic segmentation model, CM-Net, which combines the center point of the spine and the modified U-Net network. As a result, the average Dice Similarity Coefficient(DSC) was 0.974, precision was 0.916, recall was 0.906, accuracy was 0.998, and Area under the Precision-Recall Curve (AUPRC) was 0.912. This study demonstrates a high-performance automatic segmentation model for lumbar X-ray images, which overcomes noise such as spinal fractures and implants. Furthermore, we can perform accurate measurement of bone density on lumbar X-ray images using an automatic segmentation methodology for the spine, which can prevent the risk of compression fractures at an early stage and improve the accuracy and efficiency of osteoporosis diagnosis.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.7
• /
• pp.1841-1857
• /
• 2023

CenterNet, a novel object detection algorithm without anchor based on key points, regards the object as a single center point for prediction and directly regresses the object's height and width. However, because the objects have different sizes, directly regressing their height and width will make the model difficult to converge and lose the intrinsic relationship between object's width and height, thereby reducing the stability of the model and the consistency of prediction accuracy. For this problem, we proposed an algorithm based on the regression of the diagonal half-length and the center angle, which significantly compresses the solution space of the regression components and enhances the intrinsic relationship between the decoded components. First, encode the object's width and height into the diagonal half-length and the center angle, where the center angle is the angle between the diagonal and the vertical centreline. Secondly, the predicted diagonal half-length and center angle are decoded into two length components. Finally, the position of the object bounding box can be accurately obtained by combining the corresponding center point coordinates. Experiments show that, when using CenterNet as the improved baseline and resnet50 as the Backbone, the improved model achieved 81.6% and 79.7% mAP on the VOC 2007 and 2012 test sets, respectively. When using Hourglass-104 as the Backbone, the improved model achieved 43.3% mAP on the COCO 2017 test sets. Compared with CenterNet, the improved model has a faster convergence rate and significantly improved the stability and prediction accuracy.