In this paper, we propose a map alignment method for maps generated by point-line monocular SLAM. In the proposed method, the information of feature lines as well as feature points extracted from multiple maps are fused into a single map. To this end, the proposed method first searches for similar areas between maps via Bag-of-Words-based image matching. Thereafter, it calculates the similarity transformation between the maps in the corresponding areas to align the maps. Finally, we merge the overlapped information of multiple maps into a single map by removing duplicate information from similar areas. Experimental results show that maps created by different users are combined into a single map, and the accuracy of the fused map is similar with the one generated by a single user. We expect that the proposed method can be utilized for fast imagery map generation.

Recently, deep-learning based methods for low-light image enhancement accomplish great success through supervised learning. However, they still suffer from the lack of sufficient training data due to difficulty of obtaining a large amount of low-/normal-light image pairs in real environments. In this paper, we propose an unsupervised learning approach for single low-light image enhancement using the bright channel prior (BCP), which gives the constraint that the brightest pixel in a small patch is likely to be close to 1. With this prior, pseudo ground-truth is first generated to establish an unsupervised loss function. The proposed enhancement network is then trained using the proposed unsupervised loss function. To the best of our knowledge, this is the first attempt that performs a low-light image enhancement through unsupervised learning. In addition, we introduce a self-attention map for preserving image details and naturalness in the enhanced result. We validate the proposed method on various public datasets, demonstrating that our method achieves competitive performance over state-of-the-arts.

Negative emotion causes stress and lack of attention concentration. The classification of negative emotion is important to recognize risk factors. To classify emotion status, various methods such as questionnaires and interview are used and it could be changed by personal thinking. To solve the problem, we acquire multi modal bio-signals such as electrocardiogram (ECG), skin temperature (ST), galvanic skin response (GSR) and extract features. The neural network (NN), the deep neural network (DNN), and the deep belief network (DBN) is designed using the multi modal bio-signals to analyze emotion status. As a result, the DBN based on features extracted from ECG, ST and GSR shows the highest accuracy (93.8%). It is 5.7% higher than compared to the NN and 1.4% higher than compared to the DNN. It shows 12.2% higher accuracy than using only single bio-signal (GSR). The multi modal bio-signal acquisition and the deep learning classifier play an important role to classify emotion.

In this paper, we propose a smart thermostat temperature set-point control method based on machine learning and rule engine, which controls thermostat's temperature set-point so that it can achieve energy savings as much as possible without sacrifice of occupants' comfort while users' preference usage pattern is respected. First, the proposed method periodically mines data about how user likes for heating (winter)/cooling (summer) his or her home by learning his or her usage pattern of setting temperature set-point of the thermostat during the past several weeks. Then, from this learning, the proposed method establishes a weekly schedule about temperature setting. Next, by referring to thermal comfort chart by ASHRAE, it makes rules about how to adjust temperature set-points as much as low (winter) or high (summer) while the newly adjusted temperature set-point satisfies thermal comfort zone for predicted humidity. In order to make rules work on time or events, we adopt rule engine so that it can achieve energy savings properly without sacrifice of occupants' comfort. Through experiments, it is shown that the proposed smart thermostat temperature set-point control method can achieve better energy savings while keeping human comfort compared to other conventional thermostat.

Multi-view images, which are widely used for providing free-viewpoint services, can enhance the quality of synthetic views when the number of views increases. However, there needs an efficient representation method because of the tremendous amount of data. In this paper, we propose a method for generating point cloud data for the efficient representation of multi-view color and depth images. The proposed method conducts sequential reconstruction of point clouds at each viewpoint as a method of deleting duplicate data. A 3D point of a point cloud is projected to a frame to be reconstructed, and the color and depth of the 3D point is compared with the pixel where it is projected. When the 3D point and the pixel are similar enough, then the pixel is not used for generating a 3D point. In this way, we can reduce the number of reconstructed 3D points. Experimental results show that the propose method generates a point cloud which can generate multi-view images while minimizing the number of 3D points.

Leukemia induced death has been listed in the top ten most dangerous mortality basis for human being. Some of the reason is due to slow decision-making process which caused suitable medical treatment cannot be applied on time. Therefore, good clinical decision support for acute leukemia type classification has become a necessity. In this paper, the author proposed a novel approach to perform acute leukemia type classification using sequential neural network classifier. Our experimental result only cover the first classification process which shows an excellent performance in differentiating normal and abnormal cells. Further development is needed to prove the effectiveness of second neural network classifier.

Image-based Virtual Try-On Systems are among the most potential solution for virtual fitting which tries on a target clothes into a model person image and thus have attracted considerable research efforts. In many cases, current solutions for those fails in achieving naturally looking virtual fitted image where a target clothes is transferred into the body area of a model person of any shape and pose while keeping clothes context like texture, text, logo without distortion and artifacts. In this paper, we propose a new improved image-based virtual try-on network system based on keypoints, which we name as KP-VTON. The proposed KP-VTON first detects keypoints in the target clothes and reliably predicts keypoints in the clothes of a model person image by utilizing a dense human pose estimation. Then, through TPS transformation calculated by utilizing the keypoints as control points, the warped target clothes image, which is matched into the body area for wearing the target clothes, is obtained. Finally, a new try-on module adopting Attention U-Net is applied to handle more detailed synthesis of virtual fitted image. Extensive experiments on a well-known dataset show that the proposed KP-VTON performs better the state-of-the-art virtual try-on systems.

People who have sleep disorders such as insomnia take a long time to get to sleep, namely sleep latency. In order to reduce it, effective stimulations and environments to induce sleep such as ASMR or pink noise are necessary. However these have different effects and preferences for each individual. Therefore customized service and control for the sleep induction will be provide to him/her. In this paper, we proposed SIS control system which provides selectively sound control among various kinds of ASMR and pink noise according to sleep state measured from brain wave data for an individual. In order to verify the effectiveness of the system, we had conducted totally 30 experiments for 5 people, and all EEG data measured from all the people during sleep. An average of 3.7 hours was spent per experiment. In comparison experiments with and without sound control for sleep induction, the latency time was reduced by an average of 8 minutes as well as delta waves and theta waves, which appear only in deep sleep, are increased by 21%.

In this paper, a stacked convolutional autoencoder model is proposed in order to classify Alzheimer's disease with high accuracy in PET/CT images. The proposed model makes use of the latent space representation - which is also called the bottleneck, of the encoder-decoder architecture: The input image is sent through the pipeline and the encoder part, using stacked convolutional filters, extracts the most useful information. This information is in the bottleneck, which then uses Softmax classification operation to classify between Alzheimer's disease, Mild Cognitive Impairment, and Normal Control. Using the data from Dong-A University, the model performs classification in detecting Alzheimer's disease up to 98.54% accuracy.

This paper proposes a novel CTC technology, CB-sense. The CB-sense guarantees a dedicated period for the CTC that results in low duty cycle of the receiver device. In addition, it send a lot of information explosively during the dedicated CTC period. Thus, it can achieve high throughput. The CB-sense uses Clear to Send (CTS) packets to transmit information bits to the heterogeneous wireless technologies. The CTS packets block the neighboring node transmissions, so it reduces co-channel interference. In experiment results, CB-sense represents 20 times more throughput than FreeBee and below 5% symbol error rate in the interference environment.

Hadoop is a framework to process large data sets in a distributed way across clusters of nodes. It has been a popular platform to process big data, but in recent years, other platforms became competitive ones depending on the characteristics of the application. Spark is one of distributed platforms to enable real-time data processing and improve overall processing performance over Hadoop by introducing in-memory processing instead of disk I/O. Whereas Hadoop is designed to work on Java and data analysis is processed using Java API, Spark provides a variety of APIs with Scala, Python, Java and R. In this paper, the goal is to find out whether the APIs of different programming languages af ect the performances in Spark. We chose two popular APIs: Python and Scala. Python is easy to learn and is used in AI domain in a wide range. Scala is a programming language with advantages of parallelism. Our experiment shows much faster processing with Scala API than Python API. For the performance issues on AI-based analysis, further study is needed.

The recent Internet environment is characterized by the explosion of certain types of data, as the data that people want is affected by certain issues. In this paper, we propose a load balancing technique that considers the data generation forms. The concept of this technique is to prioritize some type of data when it suddenly explodes. This is a technique to build an add-on middle box on a switch to monitor packets and give priority to a queue for load balancing. This technique worked when certain types of data exploded. SDN(Software Defined Networking) has the advantage of efficiently managing a number of network equipment. However, load balancing in the SDN environment has not been studied much. Applying the proposed load balancing technique in the SDN environment can save time and budget and easily implement our policies. When the proposed load balancing technique is applied to the SDN environment, it has been found that the techniques we want can be easily applied to the network systems, and that efficient data processing is possible when certain types of data explosion.

As the accidents of cargo and passenger carrying ships occur, the stress faced by the vessel users is increasing. The risk factors for accident deterioration in remote marine area are stronger than those in road accidents, and therefore the need for safety control is greatly emphasized. To overcome such issue, a lot of effort has been made in terms of conducting research and development of the passengers and freights location tracking system. However, the inner bulkhead structures are made of aluminum/steel which is a difficult material to transfer location information through. This study aims to measure the receiving rate of LoRaWAN signal inside the aluminum/steel bulkhead structures with variation in reception strength and distance. The test was taken place at platforms with 6T and 8T enclosures to reflect the same conditions found in fishery patrol boats. The receiving rate of through-the-wall communication at 10M distance turned out to be 86% in case of using two 6T enclosures, and 73% in case of using one 8T enclosure. The result showed that the receiving rate was more affected by the thickness rather than the number of enclosures.

This paper analyzes the existing copyright management and copyright sharing model and discusses the limitations. It then proposes a consortium Blockchain-based copyright management model in which the service platform participates as a node, and discusses how to combine the My Data concept with Blockchain and smart contracts. Also, Blockchain-based CP-ABE is introduced and applied to the proposed model as a way for users to define access policies and store copyright data in encrypted form on the storage of the online service providers (OSP). Compared with the existing copyright management model, the proposed model allows the copyright holder to focus on copyright registration, license content design, and sharing, as the data subject. And it is expected to be able to transparently manage the usage records and the basis for the settlement of the copyrighted data that are shared and used on each platform.

The use of information systems improves a company's service quality and provides efficient information management. The establishment of an information system has become an essential element. In designing and developing a large number of screens when building an information system, As the number of screens increases, so does the cost and time. In addition, modifications can increase the cost of maintenance and exceed the cost of development. In order to solve these problems, there have been studies to design the screen simply and easily, but there is a shortage in reducing the maintenance cost. In this paper, we have studied the layout algorithm for managing the attributes of the screen components as data structures based on the designed screen and arranging the components on the screen. And we compared the proposed method and the methods that have been studied in the related studies. Screen component layout methods can automatically deploy screen components to consistently develop a large number of screens and reduce cost and manpower by reducing the time it takes to deploy screens.

In order to minimize casualties from marine vessel accidents that occur frequently at home and abroad, it is important to ensure the safety of the passengers aboard the vessel in the event of an accident. There is an EPIRB system as a system for disaster preparedness in the marine situation currently on the market, but there is a problem that the price is very expensive. In order to overcome the cost problem, which is a disadvantage of previous system, LoRaWAN-based communication is used. LoRaWAN communication-based vessel positioning and risk detection system based on LoRaWAN communication transmits measurement data of each module using two Beacon and GPS modules to stably perform position measurement for both indoor and outdoor situations. The rider danger situation detection system can detect the safety status of the rider using the 3-axis acceleration sensor, collect data from the rider positioning system and the rider safety status detection system, and send to server using LoRa communication. When conducting communication experiments in the long-distance maritime situation and actual communication experiments using the implemented system, it was found that the two experiments showed over 90% communication success rate on average.

Since the communication networks like the Internet collapses at disaster and calamity sites, a maintenance system that can be operated offline is required for the maintenance of various facilities. In this paper, we propose a system that memory tags attached on the facilities may transmit the emergency manual to a smart-phone, and the smart phone displays it off-line. The main issue is to design low energy mode memory tags. This study presents two kinds of methods and analyzes each's energy consumption mode. The first one is to develop memory tags by using one chip, and the next one is to design memory tags by forming multi-modules. Both ways show proper application fields under the low energy mode. This research selects the off-line maintenance system by using one chip design, and proposes the direction of contents for enhancing the effectiveness of the system. And we expect that this memory tags will be valuable for disaster scenes as well as battle fields.

Algorithm journalism refers to the practices of automated news generation using algorithms that generate human sounding narratives. Algorithm journalism is known to have strengths in automating repetitive tasks through rapid and accurate analysis of data, and has been actively used in news domains such as sports and finance. In this paper, we propose an interactive card news system that generates personalized local election articles in 2018. The system consists of modules that collects and analyzes election data, generates texts and images, and allows users to specify their interests in the local elections. When a user selects interested regions, election types, candidate names, and political parties, the system generates card news according to their interest. In the study, we examined how personalized card news are evaluated in comparison with text and card news articles by human journalists, and derived implications on the potential use of algorithm in reporting political events.

R&D(Research and Development) efficiency analysis is a very important issue in academia and industry. Although many studies have been conducted to analyze R&D(Research and Development) efficiency since the past, studies that analyzed R&D(Research and Development) efficiency considering both patentability and patent quality efficiency according to the financial performance of a company do not seem to have been actively conducted. In this study, measuring the patent application and patent quality efficiency according to financial performance, patent quality efficiency according to patent application were applied to corporate groups related to artificial intelligence hardware technology defined as GPU(Graphics Processing Unit), FPGA(Field Programmable Gate Array), ASIC(Application Specific Integrated Circuit) and Neuromorphic. We analyze the efficiency empirically and use Data Envelopment Analysis as a measure of efficiency. This study examines which companies group has high R&D(Research and Development) efficiency about artificial intelligence hardware technology.

As the cloud-based ICT(Information & Communication Technology) infrastructure matures, the existing monolithic software development method is evolving into a micro-service structure based on cloud native computing. To develop and operate the services efficiently under the cloud native environment, DevOps-based application development plans through MSA(Micro Service Architecture) design based are essential. A cloud native environment is an approach to developing and running applications that take advantage of cloud computing models such as automation of source distribution, container-based virtualization, application scalability, resource efficiency, and flexible maintenance through object independence. To implement this approach, the utilization of key elements such as DevOps, continuous delivery, micro service, and containers is essential, but there are not enough previous studies on case analyses or application methods of these key elements. Therefore, in this paper, we analyze the cases of application development in cloud native environment and propose the optimized application development process and development method through small and medium-sized SI projects.

In the era of the Fourth Industrial Revolution in the digital revolution is taking place, various attempts have been made to provide various contents in a digital environment. In this paper, agent-recruitment service system based on collaborative deep learning is proposed for the intelligent head hunting service. The service system is improved from previous research [7] using collaborative deep learning for more reliable recommendation results. The Collaborative deep learning is a hybrid recommendation algorithm using "Recurrent Neural Network(RNN)" specialized for exponential calculation, "collaborative filtering" which is traditional recommendation filtering methods, and "KNN-Clustering" for similar user analysis. The proposed service system can expect more reliable recommendation results than previous research and showed high satisfaction in user survey for verification.

If various unreasonable safety standards are preemptively and effectively readjusted, the risk of accidents can be reduced. In this paper, we proposed a machine reading comprehension-based safety standard Q&A system to secure supporting technology for effective search and analysis of safety standards for integrated and systematic management of safety standards. The proposed model finds documents related to safety standard questions in the various laws and regulations, and then divides these documents into provisions. Only those provisions that are likely to contain the answer to the question are selected, and then the BERT-based machine reading comprehension model is used to find answers to questions related to safety standards. When the proposed safety standard Q&A system is applied to KorQuAD dataset, the performance of EM 40.42% and F1 55.34% are shown.

This study analyzed the factors consisting moving poster design. In order to analyze the elements of moving poster, the researcher reorganized the elements of moving posters by referring to the previous studies on printed posters, motion graphics and works of moving posters. In this study, moving poster components were classified into 'communication', 'visual and form', and 'sound'. In 'communication' section, moving posters had narrative elements, including scenes, because of the time and movement added by the poster's original function. The 'visual and form' section was classified into graphics, layouts, movement, and time. Graphics refered to various graphic objects constituting the screen such as photo, illustration, typography, color, diagram. Layout means screen layout, size, and orientation of the screen. Movement section was divided into 'subject of movement' and 'attributes of movement'. Time was classified physical time such as playing time and subjective time felt by the audience. Also, the researcher categorized 'sound' as an additional section, since most moving posters did not include sound.

In this study, the area of interest(AOI) of Soju ad poster was tracked for analysis the time to frist fixation, the average of fixation duration and count by the study indexes. As a result of the analysis, Visual attention was higher the face than the body shape of the ad model. This means "when we look at printed ads, we see picture elements first, not language one" but language elements can't be overlooked either. Also, the importance of the model role could be verified by measuring the visual attention on the Soju ad poster. Based on the results of this study, if further research on ad posters is carried out and scientific and quantitative interpretation methods are presented, it can be used as product marketing data that can be reflected in ad model selection and poster design.

In the 1970s and 1980s, space installations in films became a new approach to film research. Many film directors and theoretical scholars believe that the space setting in films, other than textual stories, is an important core of film narrative research. Early film scholars have proposed that the film space with sensory narrative mode as the core motive force can promote people's in-depth understanding of the film content. Therefore, many researchers have tried to emphasize the importance of space in films through Hitchcock's films, but have not analyzed the elements and techniques of film space performance in detail. Reviewing the Hitchcock's classic crime film, through the space set in the traditional film, we can discover the movie techniques and concepts related to space. Therefore, this study takes Hitchcock's classic crime film as an example, and through analysis of film space expression techniques, attempts to propose factors for film space creation and bring new spatial perception effects to the audience.

This research demonstrates the effects of a culture & arts treatment program for elderly women's regional community consciousness. Study subjects of this research are 18 women whose ages are 65 years old and older. (8 women participated in a culture & arts program and 10 women participated in an elderly center program.) The culture & arts program held 16 sessions where the subjects wrote letters or drew pictures. After this program, we measured the degree of regional community consciousness. The results of the study were, first, that the cultural and arts treatment program improved the regional community consciousness among the elderly woman. Second, the treatment program has improved the sub-region of the regional community consciousness among the woman's elderly. This result shows the culture & arts treatment program can be used as an effective way to enhance the consciousness of the regional community on elderly women