• Journal of Korea Water Resources Association
• /
• v.52 no.6
• /
• pp.429-440
• /
• 2019

The numbers of multi-year droughts due to climate change are increasing worldwide. Boryeong Dam, located in Chungcheongnam-do, South Korea, was also affected by a 4-year drought from 2014 to 2017. Since traditional unilateral decision making processes to alleviate drought damage have, until now, resulted in conflicts between many of the involved groups, the need for active participation from both stakeholders and policymakers is greater than before. This study introduced Shared Vision Planning, a collaborative decision making process that involves participation from various groups of stakeholders, by organizing Water Policy Council for Climate Change Adaptation in Chungcheongnam-do. A Shared Vision Planning Model was then developed with a system dynamics software by working together with relevant stakeholders to actively reflect their requests through three council meetings. Multiple simulations that included various future climate change scenarios were conducted, and future drought vulnerability analysis results of Boryeong Dam and districts, in terms of frequency, length, and magnitude, were arrived at. It was concluded that Boryeong Dam was more vulnerable to future droughts than the eight districts. While the total water deficit in the eight districts was not so significant, their water deficit in terms of spatial discordance was proved to be more problematic. In the future, possible alternatives to the model will be implemented so that stakeholders can use it to agree on a policy for possible conflict resolutions.

• Journal of Broadcast Engineering
• /
• v.23 no.3
• /
• pp.351-360
• /
• 2018

Human emotion recognition is a research topic that is receiving continuous attention in computer vision and artificial intelligence domains. This paper proposes a method for classifying human emotions through multiple neural networks based on multi-modal signals which consist of image, landmark, and audio in a wild environment. The proposed method has the following features. First, the learning performance of the image-based network is greatly improved by employing both multi-task learning and semi-supervised learning using the spatio-temporal characteristic of videos. Second, a model for converting 1-dimensional (1D) landmark information of face into two-dimensional (2D) images, is newly proposed, and a CNN-LSTM network based on the model is proposed for better emotion recognition. Third, based on an observation that audio signals are often very effective for specific emotions, we propose an audio deep learning mechanism robust to the specific emotions. Finally, so-called emotion adaptive fusion is applied to enable synergy of multiple networks. The proposed network improves emotion classification performance by appropriately integrating existing supervised learning and semi-supervised learning networks. In the fifth attempt on the given test set in the EmotiW2017 challenge, the proposed method achieved a classification accuracy of 57.12%.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.12
• /
• pp.505-518
• /
• 2023

3D point cloud semantic segmentation is a computer vision task that involves dividing the point cloud into different objects and regions by predicting the class label of each point. Existing 3D semantic segmentation models have some limitations in performing sufficient fusion of multi-modal features while ensuring both characteristics of 2D visual features extracted from RGB images and 3D geometric features extracted from point cloud. Therefore, in this paper, we propose MMCA-Net, a novel 3D semantic segmentation model using 2D-3D multi-modal features. The proposed model effectively fuses two heterogeneous 2D visual features and 3D geometric features by using an intermediate fusion strategy and a multi-modal cross attention-based fusion operation. Also, the proposed model extracts context-rich 3D geometric features from input point cloud consisting of irregularly distributed points by adopting PTv2 as 3D geometric encoder. In this paper, we conducted both quantitative and qualitative experiments with the benchmark dataset, ScanNetv2 in order to analyze the performance of the proposed model. In terms of the metric mIoU, the proposed model showed a 9.2% performance improvement over the PTv2 model using only 3D geometric features, and a 12.12% performance improvement over the MVPNet model using 2D-3D multi-modal features. As a result, we proved the effectiveness and usefulness of the proposed model.

• Journal of Positioning, Navigation, and Timing
• /
• v.1 no.1
• /
• pp.35-43
• /
• 2012

The UGV is one of the special field robot developed for mine detection, surveillance and transportation. To achieve successfully the missions of the UGV, the accurate and reliable navigation data should be provided. This paper presents design and implementation of multi-sensor-based open architecture integrated navigation for localization of UGV. The presented architecture hierarchically classifies the integrated system into four layers and data communications between layers are based on the distributed object oriented middleware. The navigation manager determines the navigation mode with the QoS information of each navigation sensor and the integrated filter performs the navigation mode-based data fusion in the filtering process. Also, all navigation variables including the filter parameters and QoS of navigation data can be modified in GUI and consequently, the user can operate the integrated navigation system more usefully. The conventional GPS/INS integrated system does not guarantee the long-term reliability of localization when GPS solution is not available by signal blockage and intentional jamming in outdoor environment. The presented integration algorithm, however, based on the adaptive federated filter structure with FDI algorithm can integrate effectively the output of multi-sensor such as 3D LADAR, vision, odometer, magnetic compass and zero velocity to enhance the accuracy of localization result in the case that GPS is unavailable. The field test was carried out with the UGV and the test results show that the presented integrated navigation system can provide more robust and accurate localization performance than the conventional GPS/INS integrated system in outdoor environments.

• Journal of Arbitration Studies
• /
• v.21 no.3
• /
• pp.55-87
• /
• 2011

This paper is to illustrate a variety of court-annexed ADR programs and vindicate its implications of court-annexed ADR in United States. It has been almost three decades since Frank Sender articulated his vision of the multi-door courthouse. The court-annexed ADR originated from the concept of multi-door court house. Professor Sander argued that the court must transform from the court that provides litigation, only one type of dispute resolution, to the multi-door courthouse which provides a variety of dispute resolution methods including a number of ADR programs. The types of court-annexed ADR on which this paper focus are court-annexed mediation, court-annexed arbitration, mini trial, early neutral evaluation(ENE), summary jury trial, rent-a-judge, and med-arb in United States. The findings of this paper is as follows. First, the ADR movement is the irreversible and dominant phenomenon in the US court. The motivation of incorporating ADR into court is to reduce the cost of court to handle the civil disputes and to eliminate the delay of litigation process in the court. At the same time, a couple of studies of ADR revealed that the ADR program satisfied users of ADR. Second, the landscape of ADR has not been fixed. In 1970's, the court-annexed arbitration has been popular. In 1980's, the diverse kinds of ADR programs were introduced into the federal court as well as state courts, such as mini trial, early neutral evaluation(ENE), summary jury trial, and court-annexed mediation. But in 2000s, the court-annexed mediation has been the dominant type of ADR in United States. Third, the each type of ADR program has its own place for the dispute resolution. Since Korean society enters into the stage in which diverse kind of disputes occur in the areas of environment, construction, medicare, etc, it is desirable to take into consideration of the introduction of ADR to dispute resolution in Korea.

• Journal of Broadcast Engineering
• /
• v.23 no.5
• /
• pp.636-641
• /
• 2018

Recently, the digital signages display not only the rectangular shapes but also the various shapes and sizes. The high-resolution large-screen display monitors have evolved to multi-vision modes in which several screens are connected to one another. In this paper, we present the structure of an atypically shaped signage system in which the ROI changes with the rotation of the multiple displays. The inclination angle of the monitor is calculated by taking the output value of the gyro sensor inter-locked with the Arduino, and an image in which the position of the four corners is varied according to the rotation angle by using the polar coordinate system. In order to display images in the multi-screen environment, multiple displays with a gyro sensor were controlled using serial communication. As the result, we have obtained the flexibly moving monitor systems with associated images fitting in them.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.1
• /
• pp.72-76
• /
• 2009

Recently, as the interest of ubiquitous computing has been increased there has been lots of research about recognizing human activities to provide services in this environment. Especially, in mobile environment, contrary to the conventional vision based recognition researches, lots of researches are sensor based recognition. In this paper we propose to recognize the user's activity with multi-modal sensors using hierarchical dynamic Bayesian networks. Dynamic Bayesian networks are trained by the OVR(One-Versus-Rest) strategy. The inferring part of this network uses less calculation cost by selecting the activity with the higher percentage of the result of a simpler Bayesian network. For the experiment, we used an accelerometer and a physiological sensor recognizing eight kinds of activities, and as a result of the experiment we gain 97.4% of accuracy recognizing the user's activity.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.6
• /
• pp.227-233
• /
• 2019

Recently, along with the recent development of deep learning technique, neural networks are achieving success in computer vision filed. Convolutional neural network have shown outstanding performance in not only for a simple image classification task, but also for tasks with high difficulty such as object segmentation and detection. However many such deep learning models are based on supervised-learning, which requires more annotation labels than image-level label. Especially image semantic segmentation model requires pixel-level annotations for training, which is very. To solve these problems, this paper proposes a weakly-supervised semantic segmentation method which requires only image level label to train network. Existing weakly-supervised learning methods have limitations in detecting only specific area of object. In this paper, on the other hand, we use multi-classifier deep learning architecture so that our model recognizes more different parts of objects. The proposed method is evaluated using VOC 2012 validation dataset.

• Current Optics and Photonics
• /
• v.1 no.3
• /
• pp.207-213
• /
• 2017

The Adaptive Front-lighting System (AFS) is a part of the active safety system, providing optimized vision to the driver during night time and other poor-sight conditions of the road by automatic adaptation of lighting to environmental and traffic conditions. Basically, an AFS provides four different modes of the passing beam as designated in an United Nations Economic Commission for Europe regulation (ECE324-R123): neutral state or country light (Class C), urban light (Class V), highway light (Class E), and adverse weather light (Class W). In this paper, we first present an optics design for an AFS system capable of producing the Class C/V/E/W patterns requiring only on/off modulation of multi-array LEDs with no need for any additional mechanical components. The AFS optics consists of two separated modules, cutoff and spread; the cutoff module lights a narrow central area with high luminous intensity, satisfying the cutoff regulation, and the spread module forms a wide spread beam of low luminous intensity. Each module consists of two major parts; the first converts a discretely positioned LED array into a full-filled area emitting light source plane, and the second projects the light source plane to a 25 m away target plane. With the combination of these two optics modules, the four beam patterns are formed by simple on/off modulation of multi-array LEDs. Then we report the development of a prototype that was demonstrated to provide the four beam patterns.

• Journal of the Institute of Convergence Signal Processing
• /
• v.13 no.1
• /
• pp.12-18
• /
• 2012

Stereo correspondence is the central problem of stereo vision. In this paper, we propose a stereo matching scheme based on a genetic algorithm using a multi-resolution method and AD-Census. The proposed approach considers the matching environment as an optimization problem and finds the disparity by using a genetic algorithm And adaptive chronosome structure using edge pixels and crossover mechanism are employed in this technique. A cost function is composes of certain constraints whice are commonly used in stereo matching. AD-Census measure is applied to reduce disparity error. To increase the efficiency of process, we apply image pyramid method to stereo matching and calculate the initial disparity map at the coarsest resolution. Then initial disparity map is propagated to the next finer resolution, interpolated and performed disparity refinement using local feature vector. We valid our method not only reduces the search time for correspondence compared with conventional GA-based method but also ensures the validity of matching.