• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.1
• /
• pp.50-56
• /
• 2020

With the rapid development of deep neural network theory and application research, the effect of face detection has been improved. However, due to the complexity of deep neural network calculation and the high complexity of the detection environment, how to detect face quickly and accurately becomes the main problem. This paper is based on the relatively simple model of the MTCNN model, using FDDB (Face Detection Dataset and Benchmark Homepage), LFW (Field Label Face) and FaceScrub public datasets as training samples. At the same time of sorting out and introducing MTCNN(Multi-Task Cascaded Convolutional Neural Network) model, it explores how to improve training speed and Increase performance at the same time. In this paper, the dynamic image pyramid technology is used to replace the traditional image pyramid technology to segment samples, and OHEM (the online hard example mine) function in MTCNN model is deleted in training, so as to improve the training speed.

• Convergence Security Journal
• /
• v.8 no.3
• /
• pp.19-24
• /
• 2008

Common Criteria is a standard to estimate safety of information protection product such as network-level firewall system and intrusion detection system. Recently, CC version is changed from CC v.2.3 to CC v.3.1. CC v.3.1 estimation methodology requires a secured dictionary accomodation preparation for information protection product. In this research, progressed CC v3 base composition product test and research about vulnerability analysis method. Further, this paper presents specific plan sorting composition style information protection product examination methodology to existing principle and detailed methodology.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.15 no.6
• /
• pp.20-26
• /
• 2001

In this paper, we design tunnel inner section displacement measurement system which is composed of potentiometer-type displacement sensors, microcontroller-based intelligent sensing head and host computer for the management system and acquisition data. Multiple node communication bus connects the intelligent sensing heads with the host computer. For safe and re1iab1e network operation we use daisy-chain configuration, termination resistor, fail-safe biasing circuit. For tole enhancement of system utilization, we use modbus protocol. The acquisition data are transmitted to host computer and managed by database. Several data request conditions and sorting conditions are provided by management software. The utilization of designed system is confirmed by experiment.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.8 no.6
• /
• pp.122-128
• /
• 1998

This paper presents compression of image sequences based on the classification of interframe difference image blocks. classification process consists of image activity classification and energy distribution classification. In the activity classification, interframe difference image blocks are classified into activity blocks and non-activity blocks using the edge detection. In the distribution classification, activity blocks are further classified into vertical blocks, horizontal blocks, and small activity blocks using the AC energy distribution features. The RBFN, trained with numerical classification results, successfully classifies difference image blocks according to image details. Image sequence compressing based on the classification of interframe difference image blocks using the RBFN shows better compression results and less training time than the classical sorting method and the MLP network.

• The Journal of Society for e-Business Studies
• /
• v.9 no.1
• /
• pp.83-104
• /
• 2004

Recently, information systems that require storage and retrieval of huge amount of data are becoming used widely. Accordingly, research efforts have been made to develop Linux cluster file systems in the SAN environment in which clients themselves can manage metadata and access data directly. Also a semi-flat directory structure based on extendible hashing has been proposed to support fast retrieval of files[1]. In this research, we have designed and implemented the semi-flat extendible hash directory under the Linux system. In order to evaluate the practicality of the directory, we have also implemented the B+-tree based directory and experimented the performance. According to the performance comparisons, the extendible hash directory has the better performance at insert, delete, and search operations. On the other hand, the B+-tree directory is better at sorting files.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.17 no.6
• /
• pp.118-123
• /
• 2008

Weakness process can be called ferrule array and epoxy filling process at connector manufacturing process, and a lot of problems happen as think as general manufacturing process at early investment. Wished to improve this and working environment mend of worker on childhood(planning phase) and problem that is happened at done ferrule array and epoxy injection by emphasis target. By ferrule sorting and Improvement of epoxy filling process, bring authoritativeness elevation of product by fraction defective decrease of product by sized work along with productivity elevation. On the other hand, working jigs are various in characteristics of optical connector manufacturing line. There have been lots of restriction in practice because the applicability of this system is only for single type model though the network should be smooth between lines. Thus, it is not only needed the recognition of necessity in industrial line but also the development of automation system arraying ferrule and filling epoxy in the manufacturing process. It is found that the present system development enhances productivity fairly and prevents industrial disaster in the optical connector manufacturing system.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.1
• /
• pp.1-20
• /
• 2018

Service composition in the Inter-Cloud raises new challenges that are caused by the different Quality of Service (QoS) requirements of the users, which are served by different geo-distributed Cloud providers. This paper aims to explore how to select and compose such services while considering how to reach high efficiency on cost and response time, low network latency, and high reliability across multiple Cloud providers. A new hybrid multi-objective evolutionary algorithm to perform the above task called LS-NSGA-II-DE is proposed, in which the differential evolution (DE) algorithm uses the adaptive mutation operator and crossover operator to replace the those of the Non-dominated Sorting Genetic Algorithm-II (NSGA-II) to get the better convergence and diversity. At the same time, a Local Search (LS) method is performed for the Non-dominated solution set F{1} in each generation to improve the distribution of the F{1}. The simulation results show that our proposed algorithm performs well in terms of the solution distribution and convergence, and in addition, the optimality ability and scalability are better compared with those of the other algorithms.

• Smart Structures and Systems
• /
• v.21 no.5
• /
• pp.669-675
• /
• 2018

Under braking forces of a freight train, there are great longitudinal structural responses of a large freight railway cable-stayed bridge. To alleviate such adverse reactions, viscous dampers are required, whose parametric selection is one of important and arduous researches. Based on the longitudinal dynamics vehicle model, responses of a cable-stayed bridge are investigated under various cases. It shows that there is a notable effect of initial braking speeds and locations of a freight train on the structural responses. Under the most unfavorable braking condition, the parameter sensitivity analyses of viscous dampers are systematically performed. Meanwhile, a mixing method called BPNN-NSGA-II, combining the Back Propagation neural network (BPNN) and Non-Dominated Sorting Genetic Algorithm With Elitist Strategy (NSGA-II), is employed to optimize parameters of viscous dampers. The result shows that: 1. the relationships between the parameters of viscous dampers and the key longitudinal responses of the bridge are high nonlinear, which are completely different from each other; 2. the longitudinal displacement of the bridge main girder significantly decreases by the optimized viscous dampers.

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 1996.06c
• /
• pp.773-779
• /
• 1996

Dried oak mushroom have complex and various visual features. Grading and sorting of dried oak mushrooms has been done by the human expert. Though actions involved in human grading looked simple, a decision making underneath the simple action comes from the result of the complex neural processing of the visual image. Through processing details involved in human visual recognition has not been fully investigated yet, it might say human can recognize objects via one of three ways such as extracting specific features or just image itself without extracting those features or in a combined manner. In most cases, extracting some special quantitative features from the camera image requires complex algorithms and processing of the gray level image requires the heavy computing load. This fact can be worse especially in dealing with nonuniform, irregular and fuzzy shaped agricultural products, resulting in poor performance because of the sensitiveness to the crisp criteria or specific ules set up by algorithms. Also restriction of the real time processing often forces to use binary segmentation but in that case some important information of the object can be lost. In this paper, the neuro net based real time recognition algorithm was proposed without extracting any visual feature but using only the directly captured raw gray images. Specially formated adaptable size of grids was proposed for the network input. The compensation of illumination was also done to accomodate the variable lighting environment. The proposed grading scheme showed very successful results.

• The Transactions of the Korea Information Processing Society
• /
• v.2 no.3
• /
• pp.433-442
• /
• 1995

This paper presents an improved VLSI implementation of a parallel sorter to achieve O(Ν) time complexity. Many fast VLSI sort algorithms have been proposed for sorting N elements in O(log Ν) time. However, most such algorithms proposed have complex network structure without considering data input and output time. They are also very difficult to expand or to use in real applications. After analyzing the chip area and time complexity of several parallel sort algorithms with overlapping data input and output time, the most effective algorithm, the rebound sort algorithm, is implemented in VLSI with some improvements. To achieve O(Ν) time complexity, an improved rebound sorter is able to sort 8 16-bits records on a chip. And it is possible to sort more than 8 records by connecting chips in a chain vertically.