• Journal of Internet Computing and Services
• /
• v.23 no.4
• /
• pp.35-43
• /
• 2022

Makeup is the most common way to improve a person's appearance. However, since makeup styles are very diverse, there are many time and cost problems for an individual to apply makeup directly to himself/herself.. Accordingly, the need for makeup automation is increasing. Makeup transfer is being studied for makeup automation. Makeup transfer is a field of applying makeup style to a face image without makeup. Makeup transfer can be divided into a traditional image processing-based method and a deep learning-based method. In particular, in deep learning-based methods, many studies based on Generative Adversarial Networks have been performed. However, both methods have disadvantages in that the resulting image is unnatural, the result of makeup conversion is not clear, and it is smeared or heavily influenced by the makeup style face image. In order to express the clear boundary of makeup and to alleviate the influence of makeup style facial images, this study divides the makeup area and calculates the loss function using HoG (Histogram of Gradient). HoG is a method of extracting image features through the size and directionality of edges present in the image. Through this, we propose a makeup transfer network that performs robust learning on edges.By comparing the image generated through the proposed model with the image generated through BeautyGAN used as the base model, it was confirmed that the performance of the model proposed in this study was superior, and the method of using facial information that can be additionally presented as a future study.

• Journal of the Korea Convergence Society
• /
• v.12 no.5
• /
• pp.23-29
• /
• 2021

Recent deep learning-based natural language processing studies are conducting research to improve performance by training large amounts of data from various sources together. However, there is a possibility that the methodology of learning by combining data from various sources into one may prevent performance improvement. In the case of machine translation, data deviation occurs due to differences in translation(liberal, literal), style(colloquial, written, formal, etc.), domains, etc. Combining these corpora into one for learning can adversely affect performance. In this paper, we propose a new Corpus Weight Balance(CWB) method that considers the balance between parallel corpora in machine translation. As a result of the experiment, the model trained with balanced corpus showed better performance than the existing model. In addition, we propose an additional corpus construction process that enables coexistence with the human translation market, which can build high-quality parallel corpus even with a monolingual corpus.

• Korean small business review
• /
• v.42 no.4
• /
• pp.251-283
• /
• 2020

The purpose of this study is to provide assistance to the establishment of related policies to improve the level of acceptance and use of smart factories for SMEs in Korea. To this end, the Unified Technology Acceptance Model (UTAUT) was extended to select additional factors that could affect the intention to accept technology, and to demonstrate this. To achieve the research objective, a questionnaire composed of 7-point Likert scales was prepared, and a survey was conducted for manufacturing-related companies. A total of 136 questionnaires were used for statistical processing. As a result of the hypothesis test, performance expectation and social influence had a positive (+) positive effect on voluntary use, but effort expectation and promotion conditions did not have a significant effect. As an extension factor, the network effect and organizational characteristics had a positive (+) effect, and the innovation resistance had a negative effect (-), but the perceived risk had no significant effect. When the size of the company is large, the perceived risk and innovation resistance are low, and the level of influencing factors for veterinary intentions, veterinary intentions, and veterinary behaviors are excluded. Through this study, factors that could have a positive and negative effect on the adoption (reduction) of smart factory-related technologies were identified and factors to be improved and factors to be reduced were suggested. As a result, this study suggests that smart factory-related technologies should be accepted.

• Journal of Broadcast Engineering
• /
• v.26 no.6
• /
• pp.692-703
• /
• 2021

Recently, with the development of computer graphics technology, research on technology for expressing real objects as more realistic virtual graphics is being actively conducted. Point cloud is a technology that uses numerous points, including 2D spatial coordinates and color information, to represent 3D objects, and they require huge data storage and high-performance computing devices to provide various services. Video-based Point Cloud Compression (V-PCC) technology is currently being studied by the international standard organization MPEG, which is a projection based method that projects point cloud into 2D plane, and then compresses them using 2D video codecs. V-PCC technology compresses point cloud objects using 2D images such as Occupancy map, Geometry image, Attribute image, and other auxiliary information that includes the relationship between 2D plane and 3D space. When increasing the density of point cloud or expanding an object, 3D calculation is generally used, but there are limitations in that the calculation method is complicated, requires a lot of time, and it is difficult to determine the correct location of a new point. This paper proposes a method to generate additional points at more accurate locations with less computation by applying 2D interpolation to the image on which the point cloud is projected, in the V-PCC technology.

• Journal of Korea Society of Industrial Information Systems
• /
• v.27 no.6
• /
• pp.13-23
• /
• 2022

In this paper, we propose a detailed component image classification algorithm by fashion item for unstructured data retrieval in the fashion field. Due to the COVID-19 environment, AI-based online shopping malls are increasing recently. However, there is a limit to accurate unstructured data search with existing keyword search and personalized style recommendations based on user surfing behavior. In this study, pre-processing using Mask R-CNN was conducted using images crawled from online shopping sites and then classified components for each fashion item through CNN. We obtain the accuaracy for collar of the shirt's as 93.28%, the pattern of the shirt as 98.10%, the 3 classese fit of the jeans as 91.73%, And, we further obtained one for the 4 classes fit of jeans as 81.59% and the color of the jeans as 93.91%. At the results for the decorated items, we also obtained the accuract of the washing of the jeans as 91.20% and the demage of jeans accuaracy as 92.96%.

• Journal of the Korea Computer Graphics Society
• /
• v.28 no.3
• /
• pp.91-100
• /
• 2022

This study proposes the metaverse content production pipeline using ZEPETO World, one of the representative metaverse platforms in Korea. Based on the Unity 3D engine, the ZEPETO world is configured using the ZEPETO template, and the core functions of the metaverse content that enable multi-user participation such as logic, interaction, and property control are implemented through the ZEPETO script. This study utilizes the basic functions such as properties, events, and components of the ZEPETO script as well as the ZEPETO player which includes avatar loading, character movement, and camera control functions. In addition, based on ZEPETO's properties such as World Multiplayer and Client Starter, it summarizes the core synchronization process required for multiplay metaverse content production, such as object transformation, dynamic object creation, property addition, and real-time property control. Based on this, we check the proposed production pipeline by directly producing multiplay metaverse content using ZEPETO World.

• Journal of the Korean Applied Science and Technology
• /
• v.39 no.1
• /
• pp.34-41
• /
• 2022

As the use of chemical products increases in daily life, the removal of dye waste has also emerged as an important environmental issue. This dye waste can be decomposed using a photocatalyst, and the photocatalyst can be synthesized very cost-effectively by using the sol-gel technology. The sol-gel technology is not only very useful for nanoscale film formation, but also can simply form multilayer structures. Using a multiple spin coating method, in this study, a ZnO film with a multilayered structure (3 layers, 5 layers) was formed by using zinc oxide (ZnO), which is effective in decomposing various dyes. For performance comparison, a ZnO film having a single layer structure by a single spin coating method was prepared as a control. Structural and elemental analysis of ZnO film was performed using an X-ray diffraction analyzer and an energy dispersive X-ray spectrometer. A nanowire-like surface morphology could be observed through a scanning electron microscope. Additionally, UV-Vis spectrophotometer was used to measure the absorbance of UV light. The ZnO film with a five-layer structure degraded the simulated methylene blue by 49% more than the ZnO film with a single-layer structure. In conclusion, it was found that ZnO having a multilayered structure is useful as a photocatalyst that decomposes methylene blue dye more effectively.

• Journal of Intelligence and Information Systems
• /
• v.28 no.4
• /
• pp.309-328
• /
• 2022

This study empirically analyzed a Korean pre-trained language models (PLMs) designed for natural language generation. The performance of two PLMs - BART and GPT - at the task of abstractive text summarization was compared. To investigate how performance depends on the characteristics of the inference data, ten different document types, containing six types of informational content and creation content, were considered. It was found that BART (which can both generate and understand natural language) performed better than GPT (which can only generate). Upon more detailed examination of the effect of inference data characteristics, the performance of GPT was found to be proportional to the length of the input text. However, even for the longest documents (with optimal GPT performance), BART still out-performed GPT, suggesting that the greatest influence on downstream performance is not the size of the training data or PLMs parameters but the structural suitability of the PLMs for the applied downstream task. The performance of different PLMs was also compared through analyzing parts of speech (POS) shares. BART's performance was inversely related to the proportion of prefixes, adjectives, adverbs and verbs but positively related to that of nouns. This result emphasizes the importance of taking the inference data's characteristics into account when fine-tuning a PLMs for its intended downstream task.

• Journal of the Korean Society of Radiology
• /
• v.16 no.4
• /
• pp.453-462
• /
• 2022

In this paper, the artificial intelligence (AI) technology used in the medical image analysis field was analyzed through a literature review. Literature searches were conducted on PubMed, ResearchGate, Google and Cochrane Review using the key word. Through literature search, 114 abstracts were searched, and 98 abstracts were reviewed, excluding 16 duplicates. In the reviewed literature, AI is applied in classification, localization, disease detection, disease segmentation, and fit degree of registration images. In machine learning (ML), prior feature extraction and inputting the extracted feature values into the neural network have disappeared. Instead, it appears that the neural network is changing to a deep learning (DL) method with multiple hidden layers. The reason is thought to be that feature extraction is processed in the DL process due to the increase in the amount of memory of the computer, the improvement of the calculation speed, and the construction of big data. In order to apply the analysis of medical images using AI to medical care, the role of physicians is important. Physicians must be able to interpret and analyze the predictions of AI algorithms. Additional medical education and professional development for existing physicians is needed to understand AI. Also, it seems that a revised curriculum for learners in medical school is needed.

• Journal of Navigation and Port Research
• /
• v.46 no.4
• /
• pp.321-330
• /
• 2022

Busan New Port manages the largest volume of traffic among Korean ports, and accounts for 68.5% of the total volume of the Busan port. Due to this increase in volume, ultra large container ships call at Busan New Port. When the additional south container terminal as well as ongoing construction project of the west container terminal are completed, various encounters may occur at the Busan New Port entrance, which may cause collision risk.s Thus, the purpose of this study was to provide a plan to improve the safety of vessel traffic, in the in/out bound fairway of Busan New Port. For this purpose, the status of arrivals and departures of vessels in Busan New Port, was examined through maritime traffic flow analysis. Additionally, risk factors and safety measures were identified, by AHP analysis with ship operators of the study area. Also, based on the derived safety measures, scenarios were set using the Environmental Stress model (ES model), and the traffic risk level of each safety measure was identified through simulation. As a result, it is expected that setting the no entry area for one-way traffic would have a significant effect on mitigating risks at the Busan New Port entrance. This study can serve as a basis for preparing safety measures, to improve the navigation of vessels using Busan New Port. If safety measures are prepared in the future, it is necessary to verify the safety by using the traffic volume and flow changes according to the newly-opened berths.