• Journal of Electrical Engineering and Technology
• /
• v.13 no.6
• /
• pp.2468-2478
• /
• 2018

We propose the Sparse Feature Convolutional Neural Network (SFCNN) to reduce the volume of convolutional neural networks (CNNs). Despite the superior classification performance of CNNs, their enormous network volume requires high computational cost and long processing time, making real-time applications such as online-training difficult. We propose an advanced network that reduces the volume of conventional CNNs by producing a region-based sparse feature map. To produce the sparse feature map, two complementary region-based value extraction methods, cluster max extraction and local value extraction, are proposed. Cluster max is selected as the main function based on experimental results. To evaluate SFCNN, we conduct an experiment with two conventional CNNs. The network trains 59 times faster and tests 81 times faster than the VGG network, with a 1.2% loss of accuracy in multi-class classification using the Caltech101 dataset. In vehicle classification using the GTI Vehicle Image Database, the network trains 88 times faster and tests 94 times faster than the conventional CNNs, with a 0.1% loss of accuracy.

• Proceedings of the KIEE Conference
• /
• 2000.07a
• /
• pp.212-214
• /
• 2000

The distance relay technique used for transmission line protection operates overreach and underreach to the self protection region because the power system becomes complex and fault conditions are different. To solve these problems, this paper describes new technique to set the reliable self protection lesion. The trip region of the quadrilateral distance relay is set by training of multi layer recurrent elman network. The proposed network is able to reach the trip zone for the fault impedance, fault initial angle and source impedance variance correctly.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.102.3-102
• /
• 2001

In this paper, we propose a novel method that can detect face region effectively with fuzzy theory and neural network We make fuzzy rules and membership functions to describe the face color. In this algorithm, we use a perceptually uniform color space to increase the accuracy and stableness of the nonlinear color information. We use this model to extract the face candidate, and then scan it with the pre-built sliding window by using a neural network-based pattern-matching method to find eye. A neural network examines small windows of face candidate, and decides whether each window contains eye. We can standardize the face candidate geometrically with detected eyes.

• Proceedings of the KIEE Conference
• /
• 2004.05a
• /
• pp.105-107
• /
• 2004

In this work, we suggested the method which improves the efficiency of the face detection algorithm using Pulse-Coupled Neural Network. Face detection algorithm which uses the color information is independent on size, angle, and obstruction of a face. But the use of color information encounters some problems arising from skin-tone color in the background, intensity variation within faces, and presence of random noise, and so on. Depending on these conditions, we obtained the mean and variance of skin-tone colors by experiments. Then we introduce a preprocess that the pixel with a mean value of skin-tone colors has highest level value(255) and the other pixels in the skin-tone region have values between 0 and 255 according to a normal distribution with a variance. This preprocess leads to an easy decision of the linking parameters.

• Journal of the Korean association of regional geographers
• /
• v.11 no.6
• /
• pp.543-558
• /
• 2005

This paper is to research the present conditions and characteristics of the machinery and metal industries in Busan region from industrial cluster and network point of view. The result of this study is as follows. First, a independent machinery and metal industrial cluster has been shaped in Busan region although it is not so large. Second, the technological level and the competitiveness of the machinery and metal industries in Busan region is still low. The companies in Busan area are interacted by local network with low-tech and low competitiveness. Third, we need to invite the technological leading companies to Busan region for this industrial cluster's technological competitiveness is raised.

• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 2006.07a
• /
• pp.555-555
• /
• 2006

• The Journal of Korean Institute for Practical Engineering Education
• /
• v.3 no.1
• /
• pp.167-175
• /
• 2011

The purpose of the study was to examine the case of Regional Learning Network and present the implications for Consortium training Programs for small and middle-sized corporations in Korea. I found a following findings in Alesund regional learning network. First, companies participated regional learning network on their own initiative. Second, companies developed their innovation competency and had a chance to implement change management programs in order to cope with dynamic environment. Third, regional learning network was designed based on organization learning theory which make it possible to create, experiment and share the common knowledge for paticipating companies. Based on the these findings, implications for Consortium training Programs for small and middle-sized corporations in Korea was presented.

• Journal of Korean Society of Rural Planning
• /
• v.23 no.3
• /
• pp.1-12
• /
• 2017

The purpose of this study is to reveal network potential among agricultural villages focused on landscape and amenity resources. For this study, we conducted Social Network Analysis (SNA) utilizing existing landscape resource database. As a result of the study, major landscape types shared among villages were found for each city. For example, agricultural and residential landscapes were identified as major types for Danjin city. Add to major landscape resources, in Dangjin city, Habduk village were recognized as a core. Seokmun, Daehoji, Woogang, and Sunseong villages were widely found as the sub core group. For Seosan city, Jigok, Palbong, and Kobuk villages were widely recognized as core group. Most of villages which indicated the highest degree centrality were superior in terms of the number of total landscape resources as well as landscape type diversity. These results can be useful for initial planning process when considering major theme for landscape-based network organization. Also, this information will be helpful for planning stage through the specification of the potential role of each village in overall network.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.4
• /
• pp.1795-1811
• /
• 2019

In order to achieve rapid and accurate detection of vehicle objects in complex traffic conditions, we propose a novel vehicle detection method. Firstly, more contextual and small-object vehicle information can be obtained by our Joint Feature Network (JFN). Secondly, our Evolved Region Proposal Network (EPRN) generates initial anchor boxes by adding an improved version of the region proposal network in this network, and at the same time filters out a large number of false vehicle boxes by soft-Non Maximum Suppression (NMS). Then, our Mask Network (MaskN) generates an example that includes the vehicle occlusion, the generator and discriminator can learn from each other in order to further improve the vehicle object detection capability. Finally, these candidate vehicle detection boxes are optimized to obtain the final vehicle detection boxes by the Fine-Tuning Network(FTN). Through the evaluation experiment on the DETRAC benchmark dataset, we find that in terms of mAP, our method exceeds Faster-RCNN by 11.15%, YOLO by 11.88%, and EB by 1.64%. Besides, our algorithm also has achieved top2 comaring with MS-CNN, YOLO-v3, RefineNet, RetinaNet, Faster-rcnn, DSSD and YOLO-v2 of vehicle category in KITTI dataset.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.44 no.1
• /
• pp.56-63
• /
• 2007

In this paper, an eye tracking method is presented using a neural network (NN) and mean-shift algorithm that can accurately detect and track user's eyes under the cluttered background. In the proposed method, to deal with the rigid head motion, the facial region is first obtained using skin-color model and con-nected-component analysis. Thereafter the eye regions are localized using neural network (NN)-based tex-ture classifier that discriminates the facial region into eye class and non-eye class, which enables our method to accurately detect users' eyes even if they put on glasses. Once the eye region is localized, they are continuously and correctly tracking by mean-shift algorithm. To assess the validity of the proposed method, it is applied to the interface system using eye movement and is tested with a group of 25 users through playing a 'aligns games.' The results show that the system process more than 30 frames/sec on PC for the $320{\times}240$ size input image and supply a user-friendly and convenient access to a computer in real-time operation.