• Journal of Internet Computing and Services
• /
• v.20 no.4
• /
• pp.91-102
• /
• 2019

In autonomous driving systems, the ability to classify pedestrians in images captured by cameras is very important for pedestrian safety. In the past, after extracting features of pedestrians with HOG(Histogram of Oriented Gradients) or SIFT(Scale-Invariant Feature Transform), people classified them using SVM(Support Vector Machine). However, extracting pedestrian characteristics in such a handcrafted manner has many limitations. Therefore, this paper proposes a method to classify pedestrians reliably and effectively using CNN's(Convolutional Neural Network) deep features and transfer learning. We have experimented with both the fixed feature extractor and the fine-tuning methods, which are two representative transfer learning techniques. Particularly, in the fine-tuning method, we have added a new scheme, called M-Fine(Modified Fine-tuning), which divideslayers into transferred parts and non-transferred parts in three different sizes, and adjusts weights only for layers belonging to non-transferred parts. Experiments on INRIA Person data set with five CNN models(VGGNet, DenseNet, Inception V3, Xception, and MobileNet) showed that CNN's deep features perform better than handcrafted features such as HOG and SIFT, and that the accuracy of Xception (threshold = 0.5) isthe highest at 99.61%. MobileNet, which achieved similar performance to Xception and learned 80% fewer parameters, was the best in terms of efficiency. Among the three transfer learning schemes tested above, the performance of the fine-tuning method was the best. The performance of the M-Fine method was comparable to or slightly lower than that of the fine-tuningmethod, but higher than that of the fixed feature extractor method.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.7
• /
• pp.1517-1524
• /
• 2011

In this paper, a new document summarization method, which uses the semantic features and the pseudo relevance feedback (PRF) by using WordNet, is introduced to extract meaningful sentences relevant to a user query. The proposed method can improve the quality of document summaries because the inherent semantic of the documents are well reflected by the semantic feature from NMF. In addition, it uses the PRF by the semantic features and WordNet to reduce the semantic gap between the high level user's requirement and the low level vector representation. The experimental results demonstrate that the proposed method achieves better performance that the other methods.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.6
• /
• pp.1092-1099
• /
• 2017

In this paper, we propose a novel geographical on-demand routing protocol for a mobile ad-hoc network (MANET) that resolves the broadcast storm problem of ad-hoc on-demand distance vector (AODV) routing protocol. The candidates relaying route request (RREQ) packet are geographically restricted by the locations of the destination and a (RREQ sender. Then, the duplicate RREQs, which are generally discarded in the protocols that follow the principle of the AODV, are parsed to avoid redundant RREQ rebroadcasts and to perform passive acknowledgement with respect to a RREQ rebroadcast. The proposed routing protocol is implemented by modifying the source code of the AODV in QualNet. It is shown through QualNet simulations that the proposed routing protocol significantly improves the packet delivery ratio and the latency of the AODV in high-density and heavy traffic network scenarios.

• Nuclear Engineering and Technology
• /
• v.40 no.7
• /
• pp.607-614
• /
• 2008

Since pipes with wall-thinning defects can collapse at fluid pressure that are lower than expected, the collapse moment of wall-thinned pipes should be determined accurately for the safety of nuclear power plants. Wall-thinning defects, which are mostly found in pipe bends and elbows, are mainly caused by flow-accelerated corrosion. This lowers the failure pressure, load-carrying capacity, deformation ability, and fatigue resistance of pipe bends and elbows. This paper offers a support vector regression (SVR) model further enhanced with a fuzzy algorithm for calculation of the collapse moment and for evaluating the integrity of wall-thinned piping systems. The fuzzy support vector regression (FSVR) model is applied to numerical data obtained from finite element analyses of piping systems with wall-thinning defects. In this paper, three FSVR models are developed, respectively, for three data sets divided into extrados, intrados, and crown defects corresponding to three different defect locations. It is known that FSVR models are sufficiently accurate for an integrity evaluation of piping systems from laser or ultrasonic measurements of wall-thinning defects.

• The Journal of the Korea Contents Association
• /
• v.9 no.12
• /
• pp.592-599
• /
• 2009

In this paper, I explain the design and implementation of an automated drawing system by communicating with a user. I made a system providing an UI for virtual environmental actual robot to interwork with environment to make a drawing that I was able to read work, and to help convenience of a user in computers. The UI of a system was used the AWT(Abstract Window Toolkit) in NetBean tools. To draw images, I got image information from SVG vector files. This system can make images according to the demand of users by using real-time communication. Also this kind of drawing process can be an artists performance, not only the final image as a result of the work. So I suggest that this real-time drawing process which is operated by user’s command can be considered as an artists performance.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.21 no.12
• /
• pp.3265-3276
• /
• 1996

The studies of document recongnition have been focused mainly on Korean documents. But most of documents composed of Korean and other characters. So, in this paper, we propose the document recognition system that can recognize the multi-size, multi font and mixed type characters. We have utilized a large scale network model, "CombNET" which consists of a 4 layered network with combstructure. And we propose recognition method that can recognize characters without discrimination of character type. The first layer constitutes a Kohonen's SOFM network which quantizes an input feature vector space into several sub-spaces and the following 2-4 layers constitutes BP network modules which classify input data in each sub-space into specified catagories. An experimental result demonstrated the usefulness of this approach with the recognition rates of 95.6% for the training data. For the mixed type character documents we obtained the recognition rates of 92.6% and recognition speed of 10.3 characters per second.

• Development and Reproduction
• /
• v.14 no.2
• /
• pp.65-74
• /
• 2010

Placenta has been shown to be a site of expression of several of the monoamine membrane uptake transporters. However, the correlation between the expressions of norepinephrine transporter (NET) and placental development including gynecological diseases is still unknown. To investigate the expression and functions of NET in placenta, we conducted to compare NET expression in normal and preeclamptic placenta and analyzed the function of NET in HTR8-SV/neo trophoblast cells after NET gene transfection. The expression of NET was analyzed in placental tissues from the following groups of patients (none underwent labor): 1) term normal placenta (n=15); 2) term with preeclamptic placeneta (n=15); and 3) pre-term with preeclamptic placenta (n=11) using semi-quantitative RT-PCR, immunohistochemistry, and Western blot. In order to evaluate the function of NET, NET gene plasmid and NET gene-specific siRNA were trnasfected into HTR-8/SVneo trophoblast cells for 24 hours. NET had low expression in the pre-eclamptic placenta compare with normal placenta but no difference in western blot data. NET was expressed in the trophoblasts, and the up-regulation of NET gene stimulated the invasion of HTR-8/SVneo trophoblast cells by 2.5 fold (p<0.05), whereas the NET-siRNA treatment reduced invasion rates. Also, we observed that the expression of NET induces to expression and activity of MMP-9 in HTR-8/SVneo trophoblast cells in zymography. The results suggest that the expression of NET were reduced in pre-eclampsia and should be inhibited invasion activity of trophoblasts. Therefore, these findings provide useful guidelines for the mechanisms of trophoblast invasion as well as for the basic understanding of gynecological diseases including pre-eclampsia.

• Journal of Korea Multimedia Society
• /
• v.10 no.6
• /
• pp.745-753
• /
• 2007

The wireless medium is known to be time-varying which could affect and result to a poor network's performance. As a solution, an opportunistic scheduling and power control algorithm based on IEEE 802.11 MAC protocol is proposed in this paper. The algorithm opportunistically exploits the channel condition for better network performance. Convex optimization problems were also formulated i.e. the overall transmission power of the system is minimized and the "net-utility" of he system is maximized. We have proven that an optimal transmission power vector may exist, satisfying the maximum power and SINR constraints at all receivers, thereby minimizing overall transmission power and maximizing net-utility of the system.

• Water for future
• /
• v.18 no.3
• /
• pp.235-241
• /
• 1985

The predictions of the local and regional sand transport of the offshore tidal sand banks tn the west coast of Korea, east of the Yellow San have been performed on the basis of diagrammatical method suggested by Kenyon et al(1981). There is general agreements between these predictions and the earlier predictions(Choi, 1983) of net regional sand transport paths from the numerical simulation of maximum bottom stress vector during the semidiurnal period due to m2 and m4 tides. Further detailed investigations are proposed for the study of Choongang Chontoe where the predicted net sand transport direction is to the dredged navigation channel.

• The Bulletin of The Korean Astronomical Society
• /
• v.42 no.2
• /
• pp.59.5-60
• /
• 2017

We present the relationship between vector magnetic field parameters and solar major flare occurrence rate. Based on this, we are developing a forecast model of major flare (M and X-class) occurrence rate within a day using hourly vector magnetic field data of Space-weather HMI Active Region Patch (SHARP) from May 2010 to April 2017. In order to reduce the projection effect, we use SHARP data whose longitudes are within ${\pm}60$ degrees. We consider six SHARP magnetic parameters (the total unsigned current helicity, the total photospheric magnetic free energy density, the total unsigned vertical current, the absolute value of the net current helicity, the sum of the net current emanating from each polarity, and the total unsigned magnetic flux) with high F-scores as useful predictors of flaring activity from Bobra and Couvidat (2015). We have considered two cases. In case 1, we have divided the data into two sets separated in chronological order. 75% of the data before a given day are used for setting up a flare model and 25% of the data after that day are used for test. In case 2, the data are divided into two sets every year in order to reduce the solar cycle (SC) phase effect. All magnetic parameters are divided into 100 groups to estimate the corresponding flare occurrence rates. The flare identification is determined by using LMSAL flare locations, giving more numbers of flares than the NGDC flare list. Major results are as follows. First, major flare occurrence rates are well correlated with six magnetic parameters. Second, the occurrence rate ranges from 0.001 to 1 for M and X-class flares. Third, the logarithmic values of flaring rates are well approximated by two linear equations with different slopes: steeper one at lower values and lower one at higher values. Fourth, the sum of the net current emanating from each polarity gives the minimum RMS error between observed flare rates and predicted ones. Fifth, the RMS error for case 2, which is taken to reduce SC phase effect, are smaller than those for case 1.