• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.5
• /
• pp.169-176
• /
• 2020

Recently, deep neural networks (DNNs) have brought revolutions to many mobile and embedded devices by providing human-level machine intelligence for various applications. However, high inference accuracy of such DNNs comes at high computational costs, and, hence, there have been significant efforts to reduce computational overheads of DNNs either by compressing off-the-shelf models or by designing a new small footprint DNN architecture tailored to resource constrained devices. One notable recent paradigm in designing small footprint DNN models is sharing parameters in several layers. However, in previous approaches, the parameter-sharing techniques have been applied to large deep networks, such as ResNet, that are known to have high redundancy. In this paper, we propose a parameter-sharing method for already parameter-efficient small networks such as ShuffleNetV2. In our approach, small templates are combined with small layer-specific parameters to generate weights. Our experiment results on ImageNet and CIFAR100 datasets show that our approach can reduce the size of parameters by 15%-35% of ShuffleNetV2 while achieving smaller drops in accuracies compared to previous parameter-sharing and pruning approaches. We further show that the proposed approach is efficient in terms of latency and energy consumption on modern embedded devices.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.3
• /
• pp.687-696
• /
• 2014

This paper presents a semantic object extraction method using user's stroke input, color, and depth information. It is supposed that a semantically meaningful object is surrounded with a few strokes from a user, and has similar depths all over the object. In the proposed method, deciding the region of interest (ROI) is based on the stroke input, and the semantically meaningful object is extracted by using color and depth information. Specifically, the proposed method consists of two steps. The first step is over-segmentation inside the ROI using color and depth information. The second step is semantically meaningful object extraction where over-segmented regions are classified into the object region and the background region according to the depth of each region. In the over-segmentation step, we propose a new marker extraction method where there are two propositions, i.e. an adaptive thresholding scheme to maximize the number of the segmented regions and an adaptive weighting scheme for color and depth components in computation of the morphological gradients that is required in the marker extraction. In the semantically meaningful object extraction, we classify over-segmented regions into the object region and the background region in order of the boundary regions to the inner regions, the average depth of each region being compared to the average depth of all regions classified into the object region. In experimental results, we demonstrate that the proposed method yields reasonable object extraction results.

• Cartoon and Animation Studies
• /
• s.39
• /
• pp.393-412
• /
• 2015

See and feel the emotion recognition is the image of a person variously changed according to the environment, personal disposition. Thus, the image recognition has been focused on the emotional sensibilities computer you want to control the number studies. However, existing emotional computing model is numbered and the objective is clearly insufficient measurement conditions. Thus, through quantifiable image Emotion Recognition and emotion computing, is a study of the situation requires an objective assessment scheme. In this paper, the sensitivity was represented by numbered sizes quantified according to the image recognition calculation emotion. So apply the principal attributes of the color image emotion recognition as a configuration parameter. In addition, in calculating the color sensitivity by applying a digital computing focused research. Image color emotion computing research approach is the color of emotion attribute, brightness, and saturation reflects the weighted according to importance to the emotional scores. And free-degree by applying the sensitivity point to the image sensitivity formula (X), the tone (Y-axis) is calculated as a number system. There pleasure degree (X-axis), the tension and position the position of the image point that the sensitivity of the emotional coordinate crossing (Y-axis). Image color coordinates by applying the core emotional effect of Russell (Core Affect) is based on the 16 main representatives emotion. Thus, the image recognition sensitivity and compares the number size. Depending on the magnitude of the sensitivity scores demonstrate this sensitivity must change. Compare the way the images are divided up the top five of emotion recognition emotion emotions associated with 16 representatives, and representatives analyzed the concentrated emotion sizes. Future studies are needed emotional computing method of calculation to be more similar sensibility and human emotion recognition.

• The KIPS Transactions:PartB
• /
• v.13B no.2 s.105
• /
• pp.77-82
• /
• 2006

Digital halftoning is the technique to obtain a bilevel-toned image from continuous-toned image. Among halftoning methods, the error diffusion method gives better subjective quality than other halftoning ones. But it also makes edges of objects blurred. To overcome the defect, we proposes the modified error diffusion to enhance the edges using the property that human vision perceives the local average luminance and doesn't perceive a little variation of the spatial variation. The proposed method computes a spatialactivity, which is the difference between a pixel luminance and the average of its $3{\times}3$ neighborhood pixels' Iuminance weighted according to the spatial positioning. The system also usesof edge enhancement (IEE), which is computed from the normalized spatial activitymultiplied by the average luminance. The IEE is added to the quantizer's input pixel and feeds into the halftoning quantizer. The quantizer produces the halftone image having the enhanced edge. The computer experimental results show that the proposed method produces clearer bilevel-toned images than conventional methodsand the edge of objects is preserved well. Also the performance of the preposed method is improved, compared with that of the conventional method by measuring the edge correlation and the local average accordance at some ranges of viewing distance.

• Land and Housing Review
• /
• v.4 no.3
• /
• pp.225-234
• /
• 2013

The indices to choose the object countries for developing overseas industrial park were developed and applied in this paper. The results are showing as follows. First, the Korean enterprises are branched out into total 128 countries as of the first quarter of 2010, and the 13 asian countries including China, Vietnam, Japan, and Hongkong shows the majority of precedence 20 countries among the reported during 1980-2010. Second, the 3 steps of selecting the principal region to branch out, establishing assessment indices and criteria, and choosing strategical target counties were developed to choose the countries for developing overseas industrial park. The 38 of 128 countries were selected where the GDP per capita is lower than Korea, and the local reports of incorporation during 2007-2010 are more than 10 times. Then, the 10 countries were excluded where the minimum wages during 2008-2009 are similar to Korean ($815/month). Consequently, the 28 countries including China, Vietnam, and Cambodia etc. were selected as the major target regions. Third, the indices to choose countries for developing overseas industrial park are classified into 5 categories-investment condition, labor market flexibility, potential market demand, population, changing rate of the reported number of manufacturing industry, and detailed indices for each category were selected, then the weight were given with the consideration of importance. Finally, Indonesia, Mongolia, and Uzbekistan were selected as the strategical target counties where acquire the high score in labor market flexibility and investment condition, relatively undeveloped, and friendly to Korea.

• Journal of Korea Water Resources Association
• /
• v.51 no.1
• /
• pp.81-88
• /
• 2018

The rainfall-runoff model made of sewer networks in the urban area is vast and complex, making it unsuitable for real-time urban flood forecasting. Therefore, the rainfall-runoff model is constructed and simplified using the sewer network of Daerim baisn. The network simplification process was composed of 5 steps based on cumulative drainage area and all parameters of SWMM were calculated using weighted area. Also, in order to estimate the optimal simplification range of the sewage network, runoff and flood analysis was carried out by 5 simplification ranges. As a result, the number of nodes, conduits and the simulation time were constantly reduced to 50~90% according to the simplification ranges. The runoff results of simplified models show the same result before the simplification. In the 2D flood analysis, as the simplification range increases by cumulative drainage area, the number of overflow nodes significantly decreased and the positions were changed, but similar flooding pattern was appeared. However, in the case of more than 6 ha cumulative drainage area, some inundation areas could not be occurred because of deleted nodes from upstream. As a result of comparing flood area and flood depth, it was analyzed that the flood result based on simplification range of 1 ha cumulative drainage area is most similar to the analysis result before simplification. It is expected that this study can be used as reliable data suitable for real-time urban flood forecasting by simplifying sewer network considering SWMM parameters.

• Journal of Korea Water Resources Association
• /
• v.53 no.8
• /
• pp.583-595
• /
• 2020

In this study, using the RCP scenario for Hyoja Drainage subbasin of Cheonggyecheon, we analyzed the change with the Historical and Future rainfall calculated from five GCMs models. As a result of analyzing the average rainfall by each GCMs model, the future rainfall increased by 35.30 to 208.65 mm from the historical rainfall. Future rainfall increased 1.73~16.84% than historical rainfall. In addition, the applicability of LID element technologies such as porous pavement, infiltration trench and green roof was analyzed using the SWMM model. And the applied weight and runoff for each LID element technology are analyzed. As a result of the analysis, although there was a difference for each GCMs model, the runoff increased by 2.58 to 28.78%. However, when single porous pavement and Infiltration trench were applied, Future rainfall decreased by 3.48% and 2.74%, 8.04% and 7.16% in INM-CM4 and MRI-CGCM3 models, respectively. Also, when the two types of LID element technologies were combined, the rainfall decreased by 2.74% and 2.89%, 7.16% and 7.31%, respectively. This is less than or similar to the historical rainfall runoff. As a result of applying the LID elemental technology, it was found that applying a green roof area of about 1/3 of the urban area is the most effective to secure the lag time of runoff. Moreover, when applying the LID method to the old downtown area, it is desirable to consider the priority order in the order of economic cost, maintenance, and cityscape.

• Journal of KIISE:Databases
• /
• v.34 no.3
• /
• pp.193-205
• /
• 2007

Increasing amount of digital images requires more accurate and faster way of image retrieval. So far, image retrieval method includes content-based retrieval and keyword based retrieval, the former utilizing visual features such as color and brightness and the latter utilizing keywords which describe the image. However, the effectiveness of these methods as to providing the exact images the user wanted has been under question. Hence, many researchers have been working on relevance feedback, a process in which responses from the user are given as a feedback during the retrieval session in order to define user’s need and provide improved result. Yet, the methods which have employed relevance feedback also have drawbacks since several feedbacks are necessary to have appropriate result and the feedback information can not be reused. In this paper, a novel retrieval model has been proposed which annotates an image with a keyword and modifies the confidence level of the keyword in response to the user’s feedback. In the proposed model, not only the images which have received positive feedback but also the other images with the visual features similar to the features used to distinguish the positive image are subjected to confidence modification. This enables modifying large amount of images with only a few feedbacks ultimately leading to faster and more accurate retrieval result. An experiment has been performed to verify the effectiveness of the proposed model and the result has demonstrated rapid increase in recall and precision while receiving the same number of feedbacks.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.4
• /
• pp.101-110
• /
• 2017

The National Emergency Management Agency (NEMA) presented the disaster prevention performance target rainfall (DPPTR) for disaster prevention. The estimation criteria for DPPTR is a 10 year cycle. On the other hand, the target rainfall recalculated every 10 years is difficult to reflect the current change in rainfall on climate change. In this study, the probability of precipitation using the recent rainfall data was prepared and the weights according to socio-economic criteria reflecting the urban characteristics and adjusted probability rainfall criteria were applied to the results. The difference between the existing target rainfall and recalculated result was compared. The input data for the estimated probability rainfall was selected from 6 points located in the rainfall observing station of Chungcheongnam-do, Daejeon region. As a result of the estimation, in the case of upward probability precipitation weight, some similar areas were observed. On the other hand, there were a few cases of upward or downward changes within 10 mm. Considering the rainfall variability and uncertainty due to climate change, the existing target rainfall does not present the condition properly. Therefore, hydrological designers need to calculate the target rainfall, reflecting the present condition.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.4
• /
• pp.353-359
• /
• 2021

The Army is constructing a training system using Miles equipment that applies the latest science and technology to carry out military training. The Miles training system is a system that uses Miles equipment to simulate the damage situation of combat personnel and equipment in the same way as an actual battlefield by conducting practiced maneuvers in the field. Through this, the training force can experience conditions similar to an actual battle. In particular, the training effects of the warriors participating in the training can be maximized by establishing an integrated system that utilizes cutting-edge science technologies, such as information communication and computer simulation. This study analyzed the effects of Miles training in the army using scientific techniques targeted at the mid-range Miles. In particular, the effect index for analyzing the training effect was derived from a literature survey and expert opinions. The weight of each effect index was calculated by applying the Swing method. The final training effect was calculated by combining the results of the survey from train-experienced people. The Miles training effect was 2.6 times more effective than previous training without using Miles, and the satisfaction rate with Miles training according to status was high through variance analysis, and the difference was statistically significant.