• Journal of the Korea Computer Graphics Society
• /
• v.30 no.3
• /
• pp.109-123
• /
• 2024

Research into vision-based end-to-end autonomous driving systems utilizing deep learning and reinforcement learning has been steadily increasing. These systems typically encode continuous and high-dimensional vehicle states, such as location, velocity, orientation, and sensor data, into latent features, which are then decoded into a vehicular control policy. The complexity of urban driving environments necessitates the use of state representation learning through networks like Variational Autoencoders (VAEs) or Convolutional Neural Networks (CNNs). This paper analyzes the impact of different image state encoding methods on reinforcement learning performance in autonomous driving. Experiments were conducted in the CARLA simulator using RGB images and semantically segmented images captured by the vehicle's front camera. These images were encoded using VAE and Vision Transformer (ViT) networks. The study examines how these networks influence the agents' learning outcomes and experimentally demonstrates the role of each state representation technique in enhancing the learning efficiency and decision- making capabilities of autonomous driving systems.

• Journal of the Society of Disaster Information
• /
• v.20 no.3
• /
• pp.663-671
• /
• 2024

Purpose: The purpose of this study is to examine the risks and criminal progress of pig murder fraud crimes in the absence of related research and statistics amid a surge in pig murder fraud damage in Korea, and to approach the methods, problems, and countermeasures that can be prevented in Korea through foreign cases. Method: In order to classify the types of pig murder fraud crimes, the progress was analyzed through related reports and statistics, and various damage cases. Result: The crime of pig murder fraud can be seen as an organized crime that organically combines simple romance scams, investment fraud, and human trafficking. In the case of victims, unlike simple fraud, psychological problems such as suicide impulse and panic disorder were found, and in the case of cryptocurrency acquired through fraud, a path was found to flow back into the criminal organization through money laundering. Conclusion: In the case of pig murder fraud crimes that have been detected as a danger worldwide, the scale will be even wider. Therefore, various institutional supplements and policies should be prepared through crime type analysis in Korea for crime prevention and countermeasures.

• Korean Journal of Culture and Social Issue
• /
• v.30 no.3
• /
• pp.321-348
• /
• 2024

This study examined the relationship between prosocial motivation and acquiescent, defensive, and relational silence behaviors, and identified the mediating effect of emotional job burnout. We also examined whether these relationships differed depending on whether the organization is currently working virtually or not. To this end, we collected 359 samples from Korean employees working in general office jobs below managerial positions using online survey methods. The data were analyzed using SPSS 21.0 and PROCESS Macro, and the results showed that the relationship between prosocial motivation and acquiescent, defensive, and relational silence behaviors was mediated by emotional job burnout, with a partial mediation effect for acquiescent silence and a full mediation effect for defensive and relational silence behaviors. Furthermore, these relationships were found to be the same regardless of whether employees worked virtually or not. Finally, we discuss the implications of our findings, limitations of this study, and suggestions for future research.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.10
• /
• pp.165-173
• /
• 2024

Game engines are essential tools in game development, speeding up processes and simplifying the integration of various modules like physics, graphics, animations, and AI. This study provides a comprehensive overview of modern game engine technologies, including advanced rendering techniques, graphics APIs, physics simulations, AI integration, audio systems, networking, VR/AR, and development tools. It highlights recent advancements such as real-time ray tracing, physically based rendering, machine learning for content generation and intelligent NPCs, cloud gaming, and novel input methods like brain-computer interfaces. The paper also explores future directions, including enhanced cross-platform support and new technologies that will drive the evolution of game engines. This analysis serves as a valuable resource for developers, researchers, and industry professionals.

• International conference on construction engineering and project management
• /
• 2024.07a
• /
• pp.585-594
• /
• 2024

In recent years, the construction industry has rapidly adopted offsite-manufacturing and distributed construction methods. This change brings a variety of challenges requiring innovative solutions, such as the utilization of AI-driven and generative design. Numerous studies have explored the concept of multi-objective generative design with genetic algorithms in construction. However, this paper highlights the challenges and proposes a solution for combining generative design with distributed construction to address the need for agility in design. To achieve this goal, the research delves into the development of a multi-objective generative design optimization using a weighted genetic algorithm based on simulated annealing. The specific design case adopted is an educational complex. The proposed process strives for scalable economic viability, environmental comfort, and operational efficiency by optimizing modular configurations of architectural spaces, facilitating affordable, scalable, and optimized construction. Rhino-Grasshopper and Galapagos design tools are used to create a virtual environment capable of generating architectural configurations within defined boundaries. Optimization factors include adherence to urban regulations, acoustic comfort, and sunlight exposure. A normalized scoring approach is also presented to prioritize design preferences, enabling systematic and data-driven design decision-making. Building Information Modeling (BIM) tools are also used to transform the optimization results into tangible architectural elements and visualize the outcome. The resulting process contributes both to practice and academia. Practitioners in AEC industry could gain benefit through adopting and adapting its features with the unique characteristics of various construction projects while educators and future researchers can modify and enhance this process based on new requirements.

• International conference on construction engineering and project management
• /
• 2024.07a
• /
• pp.1308-1308
• /
• 2024

Computer vision techniques have been widely employed in automated construction management to enhance safety and prevent accidents at construction sites. However, previous research in the field of vision-based approaches has often overlooked small-sized construction tools. These tools present unique challenges in data collection due to their diverse shapes and sizes, as well as in improving model performance to accurately detect and classify them. To address these challenges, this study aimed to enhance the performance of vision-based classifiers for small-sized construction tools, including bucket, cord reel, hammer, and tacker, by leveraging synthetic images generated from a 3D virtual environment. Three classifiers were developed using the YOLOv8 algorithm, each differing in the composition of the training dataset: (i) 'Real-4000', trained on 4,000 authentic images collected through web crawling methods (1,000 images per object); (ii) 'Hybrid-4000', consisting of 2,000 authentic images and 2,000 synthetic images; and (iii) 'Hybrid-8000', incorporating 4,000 authentic images and 4,000 synthetic images. To validate the performance of the classifiers, 144 directly-captured images for each object were collected from real construction sites as the test dataset. The mean Average Precision at an IoU threshold of 0.5 (mAP_0.5) for the classifiers was 79.6%, 90.8%, and 94.8%, respectively, with the 'Hybrid-8000' model demonstrating the highest performance. Notably, for objects with significant shape variations, the use of synthetic images led to the enhanced performance of the vision-based classifiers. Moreover, the practical applicability of the proposed classifiers was validated through confidence scores, particularly between the 'Hybrid-4000' and 'Hybrid-8000' models. Statistical analysis using t-tests indicated that the performance of the 'Hybrid-4000' model would either matched or exceeded that of the 'Hybrid-8000'model based on confidence scores. Thus, employing the 'Hybrid-4000' model may be preferable in terms of data collection efficiency and processing time, contributing to enhanced safety and real-time automation and robotics in construction practices.

• The Journal of the Acoustical Society of Korea
• /
• v.43 no.5
• /
• pp.547-556
• /
• 2024

To predict the response of dynamic systems through analysis, it is essential to accurately estimate the system's stiffness and apply it to the analytical model. However, directly measuring the stiffness of actual mechanical systems is challenging. Many existing methods involve decomposing the system into components, obtaining the frequency response for each component, and then reassembling them to determine the overall system response. This process can be cumbersome, and variations in coupling conditions between components can increase errors. In this study, a new method is proposed to estimate system stiffness indirectly through experiments without decomposing the system into components. The approach combines response measurements from the entire system with a theoretical model for analysis. It simplifies the stiffness source into a lumped mass model and constructs the equations of motion based on a reduced-order model of the entire system. Subsequently, the stiffness is quantified by calculating the interface forces between the stiffness source and the receiver using vibration measurements obtained at arbitrary positions through experimentation.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.7
• /
• pp.151-157
• /
• 2020

This study is to find out the effectiveness by Infant CPR method of a single rescuer. It was conducted for 51 general public. And typical infant CPR method by a single rescuer and a new method, CPR with two thumb chest compressions wrapped in both hands over the head were compared. SPSS 22.0 was used as an analysis method and to compare the both CPR methods, Paired t-test was used. As a result of the study, the average chest compression depth(39.38±1.07 mm) by CPR with two thumb chest compressions wrapped in both hands over the head was significantly high(p<0.001). Ease of mouth-to-mouth resuscitation(p<0.001), convenience of CPR method(p<0.001), and finger pain level(p<0.001) had a significant difference. As for the preference of the CPR method, 80.4%(41 people) preferred CPR with two thumb chest compressions wrapped in both hands over the head. In this study, CPR with two thumb chest compressions wrapped in both hands over the head showed more effective results than typical CPR method. However, as a virtual study using mannequins, further research is needed to apply high-quality CPR methods to field.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.49 no.12
• /
• pp.156-163
• /
• 2012

Recently, Information Hiding Technology are becoming increasingly demanding in the field of international security, military and medical image This paper proposes data hiding technique utilizing parity checker for gray level image. many researches have been adopted LSB substitution and XOR operation in the field of steganography for the low complexity, high embedding capacity and high image quality. But, LSB substitution methods are not secure through it's naive mechanism even though it achieves high embedding capacity. Proposed method replaces LSB of each pixel with XOR(between the parity check bit of other 7 MSBs and 1 Secret bit) within one pixel. As a result, stego-image(that is, steganogram) doesn't result in high image degradation. Eavesdropper couldn't easily detect the message embedding. This approach is applying the concept of symmetric-key encryption protocol onto steganography. Furthermore, 1bit of symmetric-key is generated by the self-reference of each pixel. Proposed method provide more 25% embedding rate against existing XOR operation-based methods and show the effect of the reversal rate of LSB about 2% improvement.

• The KIPS Transactions:PartB
• /
• v.13B no.3 s.106
• /
• pp.295-300
• /
• 2006

The integral imaging system is an auto-stereoscopic display that allows users to see 3D images without wearing special glasses. In the integral imaging system, the 3D object information is taken from several view points and stored as elemental images. Then, users can see a 3D reconstructed image by the elemental images displayed through a lens array. The elemental images can be created by computer graphics, which is referred to the computer-generated integral imaging. The process of creating the elemental images is called image mapping. There are some image mapping methods proposed in the past, such as PRR(Point Retracing Rendering), MVR(Multi-Viewpoint Rendering) and PGR(Parallel Group Rendering). However, they have problems with heavy rendering computations or performance barrier as the number of elemental lenses in the lens array increases. Thus, it is difficult to use them in real-time graphics applications, such as virtual reality or real-time, interactive games. In this paper, we propose a new image mapping method named VVR(Viewpoint Vector Rendering) that improves real-time rendering performance. This paper describes the concept of VVR first and the performance comparison of image mapping process with previous methods. Then, it discusses possible directions for the future improvements.