• The Journal of the Convergence on Culture Technology
• /
• v.9 no.5
• /
• pp.591-596
• /
• 2023

This study developed a program that can produce videos easily and conveniently, focusing on teenage media producers. Through user research, we identified the needs and problems of teenage producers, and implemented a character customization function desired by users and an emotion and action recommendation system using GPT. In the rendering process, the final image was created by combining audio and video using OpenCV and FFmpeg. Teenage users who do not have expertise in video production can customize web drama characters through a simple interface and receive recommendations for emotions and actions with the help of GPT. The program of this study is expected to be a tool that can help teenage users who do not have expertise in editing and directing to produce high-quality videos, lower the entry barrier to video production, and contribute to the development of the one-person media industry. do. In the future, we plan to provide a video production environment considering mobile or vertical resolution versions.

• Journal of radiological science and technology
• /
• v.46 no.5
• /
• pp.427-433
• /
• 2023

This study aimed to quantitatively compare the diameters measured directly from the coronal plane or sagittal plane of 2D digital subtraction angiography (DSA) and the cross-sectional area-converted diameters calculated from contrast-enhanced MR (CE-MR) imaging. A retrospective analysis was conducted on 20 patients who underwent both 2D DSA and CE-MR imaging. Firstly, the venous diameters of the superior sagittal sinus (SSS) and transverse sinus (TS) were directly measured from 2D DSA. Subsequently, the axial planes for SSS diameter and the sagittal plane for TS in CE-MR imaging were utilized to calculate cross-sectional area-based converted diameters. The numerical values obtained from 2D DSA and CE-MR imaging were compared pairwise at each location. For SSS, the diameter measured by 2D DSA was 27% larger than the conversion-based diameter from CE-MR imaging (9.8±1.4 mm vs. 7.1±1.3 mm, P<0.05). Similarly, for the right TS, the difference was 16% (8.8±3.2 mm vs. 7.4±2.0 mm, P<0.05), and for the left TS, the difference was 22% (8.4±2.8 mm vs. 6.6±1.3 mm, P<0.05). In conclusion, the diameter measured directly in conventional 2D DSA may be larger than the diameter converted based on the cross-sectional area. Therefore, when selecting the size of the stent, it is crucial to make precise determinations while keeping this fact in mind.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.46 no.4
• /
• pp.312-320
• /
• 2023

Recently, there has been an increasing attempt to replace defect detection inspections in the manufacturing industry using deep learning techniques. However, obtaining substantial high-quality labeled data to enhance the performance of deep learning models entails economic and temporal constraints. As a solution for this problem, semi-supervised learning, using a limited amount of labeled data, has been gaining traction. This study assesses the effectiveness of semi-supervised learning in the defect detection process of manufacturing using the MixMatch algorithm. The MixMatch algorithm incorporates three dominant paradigms in the semi-supervised field: Consistency regularization, Entropy minimization, and Generic regularization. The performance of semi-supervised learning based on the MixMatch algorithm was compared with that of supervised learning using defect image data from the metal casting process. For the experiments, the ratio of labeled data was adjusted to 5%, 10%, 25%, and 50% of the total data. At a labeled data ratio of 5%, semi-supervised learning achieved a classification accuracy of 90.19%, outperforming supervised learning by approximately 22%p. At a 10% ratio, it surpassed supervised learning by around 8%p, achieving a 92.89% accuracy. These results demonstrate that semi-supervised learning can achieve significant outcomes even with a very limited amount of labeled data, suggesting its invaluable application in real-world research and industrial settings where labeled data is limited.

• Journal of the Korean Institute of Electrical and Electronic Material Engineers
• /
• v.37 no.1
• /
• pp.79-87
• /
• 2024

This paper is a study on the optimal microdroplet generation conditions in indirect charging electrostatic spraying. Unlike the direct charging method, which applies power to the nozzle, the indirect charging method applies power to the discharge electrode between the nozzle and the collection electrode. Therefore, an electrically simplified system can be obtained by minimizing the insulation part a stable spray pattern can be obtained with a wide spray angle, and a stable spray pattern can be obtained with a wide spray angle. To conduct the study, an indirect charging type electrostatic spray visualization system was constructed and the static characteristics of the microdroplets were analyzed through image processing of the spray shape of the microdroplets. The total number of microdroplets and the number of microdroplets per power consumption are confirmed according to the changes in the distance between the discharge electrode and the collection electrode, the flow rate, and the applied voltage, which affect the generation of microdroplets, and using this, the optimal generation conditions are derived and the corresponding microdroplet size distribution was analyzed. As a result of the experiment, it was confirmed that the optimal generation condition was at a flow rate of 15 to 20 mL/min and a voltage of -22.5 to -25 kV in terms of the number of microdroplets, and at a flow rate of 15 to 20 mL/min and a voltage of -20 kV in terms of energy consumption efficiency.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.16 no.5
• /
• pp.292-300
• /
• 2023

This paper proposes a patent discovery method and strategy for Generative AI-related patents by utilizing qualitative evaluation indicators established based on the core components of the technology. Currently, the evaluation of patent quality relies on quantitative indicators, but existing quantitative indicators cannot represent the characteristics of Generative AI technology, making it difficult to accurately evaluate. Therefore, there is a need for additional qualitative indicators that consider technical characteristics based on patent claims, which can reveal the actual strength of the patent. In this paper, we propose a new evaluation index considering the technical characteristics of Generative AI. Core patents were selected using the proposed evaluation index, and the appropriateness of the proposed index was verified through the existing quantitative evaluation method for the selected core patents.

• Korean Journal of Agricultural Science
• /
• v.50 no.3
• /
• pp.513-525
• /
• 2023

This study was conducted to compare the level of vegetation recovery based on the forest restoration techniques (natural restoration and artificial restoration) determined using the satellite imagery that targeted forest fire damaged areas in Goseong-gun, Gangwon-do. The study site included the area affected by the Goseong forest fire (1996) and the East Coast forest fire (2000). We conducted a time-series analysis of satellite imagery on the natural restoration sites (19 sites) and artificial restoration sites (12 sites) that were created after the forest fire in 1996. In the analysis of satellite imagery, the difference normalized burn ratio (dNBR) and normalized difference vegetation index (NDVI) were calculated to compare the level of vegetation recovery between the two groups. We discovered that vegetation was restored at all of the study sites (31 locations). The satellite image-based analysis showed that the artificial restoration sites were relatively better than the natural restoration sites, but there was no statistically significant difference between the two groups (p > 0.05). Therefore, it is necessary to select a restoration technique that can achieve the goal of forest restoration, taking the topography and environment of the target site into account. We also believe that in the future, accurate diagnosis and analysis of the vegetation will be necessary through a field survey of the forest fire-damaged sites.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_1
• /
• pp.683-693
• /
• 2023

In this study, we used satellite images to analyze the green algae phenomenon that first occurred in Soyang-ho, which was completed in 1973. The research data used 13 optical images over a period of about 2 months from July 2023, and the area of green algae that occurred in Soyang-ho was calculated. To calculate the exact area where green algae occurred, image classification was performed based on the support vector machine algorithm. As a result, green algae in Soyang-ho occurred around the point where the impurities that caused the green algae were introduced. It seemed to temporarily decrease due to the effects of Typhoon Khanun in August 2023, but green algae increased again due to the continued heat. Soyang-ho is one of the major water sources in the metropolitan area, suggesting that we must prepare for repeated green algae outbreaks.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_1
• /
• pp.507-519
• /
• 2023

Multi-temporal optical images have been utilized for time-series monitoring of croplands. However, the presence of clouds imposes limitations on image availability, often requiring a cloud removal procedure. This study assesses the applicability of various machine learning algorithms for effective cloud removal in optical imagery. We conducted comparative experiments by focusing on two key variables that significantly influence the predictive performance of machine learning algorithms: (1) land-cover types of training data and (2) temporal variability of land-cover types. Three machine learning algorithms, including Gaussian process regression (GPR), support vector machine (SVM), and random forest (RF), were employed for the experiments using simulated cloudy images in paddy fields of Gunsan. GPR and SVM exhibited superior prediction accuracy when the training data had the same land-cover types as the cloud region, and GPR showed the best stability with respect to sampling fluctuations. In addition, RF was the least affected by the land-cover types and temporal variations of training data. These results indicate that GPR is recommended when the land-cover type and spectral characteristics of the training data are the same as those of the cloud region. On the other hand, RF should be applied when it is difficult to obtain training data with the same land-cover types as the cloud region. Therefore, the land-cover types in cloud areas should be taken into account for extracting informative training data along with selecting the optimal machine learning algorithm.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.16 no.5
• /
• pp.347-353
• /
• 2023

Measuring the growth of fish in fish farms still uses a laborious method. This method requires a lot of labor and causes stress to the fish, which has a negative impact on mortality. To solve this problem, we propose the Fish Growth Regression System (FGRS), a system to automate the growth of fish. FGRS consists of two modules. The first is a module that detects fish based on Yolo v8, and the second consists of a module that predicts the growth of fish using fish image data and a CNN-based neural network model. As a result of the simulation, the average prediction error before learning was 134.2 days, but after learning, the average error decreased to 39.8 days. It is expected that the system proposed in this paper can be used to predict the growing date and use the growth prediction of fish to contribute to automation in fish farms, resulting in a significant reduction in labor and cost savings.

• Smart Structures and Systems
• /
• v.32 no.3
• /
• pp.135-151
• /
• 2023

Using 3D asphalt pavement surface data, a deep and multiscale network named CrackNet-M is proposed in this paper for pixel-level crack detection for improvements in both accuracy and robustness. The CrackNet-M consists of four function-specific architectural modules: a central branch net (CBN), a crack map enhancement (CME) module, three pooling feature pyramids (PFP), and an output layer. The CBN maintains crack boundaries using no pooling reductions throughout all convolutional layers. The CME applies a pooling layer to enhance potential thin cracks for better continuity, consuming no data loss and attenuation when working jointly with CBN. The PFP modules implement direct down-sampling and pyramidal up-sampling with multiscale contexts specifically for the detection of thick cracks and exclusion of non-crack patterns. Finally, the output layer is optimized with a skip layer supervision technique proposed to further improve the network performance. Compared with traditional supervisions, the skip layer supervision brings about not only significant performance gains with respect to both accuracy and robustness but a faster convergence rate. CrackNet-M was trained on a total of 2,500 pixel-wise annotated 3D pavement images and finely scaled with another 200 images with full considerations on accuracy and efficiency. CrackNet-M can potentially achieve crack detection in real-time with a processing speed of 40 ms/image. The experimental results on 500 testing images demonstrate that CrackNet-M can effectively detect both thick and thin cracks from various pavement surfaces with a high level of Precision (94.28%), Recall (93.89%), and F-measure (94.04%). In addition, the proposed CrackNet-M compares favorably to other well-developed networks with respect to the detection of thin cracks as well as the removal of shoulder drop-offs.