• Science of Emotion and Sensibility
• /
• v.17 no.2
• /
• pp.55-62
• /
• 2014

In this study, a questionnaire was developed to assess error search and correction tasks, and an analysis was performed on the accuracy of the tasks and the time required for their completion in order to identify the effects of LED light source illumination on visual performance according to changes in a predicted mean vote(PMV) thermal environment. In addition, a subjective evaluation was performed by conducting a survey on the level of visual fatigue experienced during the tasks. In the experiment, four types of PMV thermal environments were established according to PMV values in the temperature range of $(17{\pm}1-29{\pm}1)^{\circ}C$ and the humidity range of $(50{\pm}5-60{\pm}5)%$, and the LED light source illumination was divided into three types: 400lx, 700lx, and 1000lx. The experimental results confirmed that the accuracy of the error search(LED p value=0.058, PMV*LED p value=0.083) and correction tasks and the time required(LED p value=0.004, PMV p value=0.000) for their completion were affected by changes in both the PMV thermal environment and the LED light source illumination, whereas a significant difference in visual fatigue was observed only in the PMV thermal environment(p value=0.003).

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 1995.10a
• /
• pp.295-298
• /
• 1995

We performed a research to improve the performance of active bar model which is used in tracking algorithm. Active bar model is a simplified model of snake model. If we used the sctive bar model, the numerical procedure for real time tracking problem can be carried out faster than snake model. However the demerit of active bar algorithms is that we can't used the provious image data because each time it has to reconstruct the active bar. In this paper we proposed advanced algorithm for active bar model. The proposed model can improve tracking abilities by preserving the active bar during the process and changing the energy functional.

• KEPCO Journal on Electric Power and Energy
• /
• v.8 no.2
• /
• pp.137-141
• /
• 2022

To inspect the power facility faults using artificial intelligence, it need that improve the accuracy of the diagnostic model are required. Data augmentation skill using generative adversarial network (GAN) is one of the best ways to improve deep learning performance. GAN model can create realistic-looking fake images using two competitive learning networks such as discriminator and generator. In this study, we intend to verify the effectiveness of virtual data generation technology by including the fake image of power facility generated through GAN in the deep learning training set. The GAN-based fake image was created for damage of LP insulator, and ResNet based normal and defect classification model was developed to verify the effect. Through this, we analyzed the model accuracy according to the ratio of normal and defective training data.

• Journal of the Korea Computer Graphics Society
• /
• v.19 no.3
• /
• pp.1-11
• /
• 2013

Fluid Implicit Particle (FLIP) method is used in Visual Effect(VFX) industries frequently because FLIP based simulation show high performance with good visual quality. However in large-scale fluid simulations, the efficiency of FLIP method is low because it requires many particles to represent large volume of water. In this papers, we propose a novel hybrid method of simulating fluids to supplement this drawback. To improve the performance of the FLIP method by reducing the number of particles, particles are deployed inside thin layers of the inner surface of water volume only. The coupling between less-disspative solutions of FLIP method and viscosity solution of level set method is achieved by introducing a new surface reconstruction method motivated by surface reconstruction method[1] and moving least squares(MLS) method[2]. Our hybrid method can generate high quality of water simulations efficiently with various multiscale features.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2012.05a
• /
• pp.797-798
• /
• 2012

The purpose of this study to embody LED visible light communication visual signal delivery system based on TFT-LCD module and to analyze a performance for digital signals. LED lighting enables to transmit data through light, to embody visible light communication, and to control lights in different wavelengths according to LED materials and structural materials. In this study, we aim to create a transmitting and receiving sensor that has implemented LED and visual sensors to LED efficiency and existing TFT-LCD module so as to estimate and indicate a communication distance and to find out application methods and problems. Also, by implementing TFT-LCD technology and next-generation LED visual communication technology, this study aims to check the possibility of technological realization by applying them to various means of communication including ubiquitous communication.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.22 no.3
• /
• pp.379-385
• /
• 2012

This paper developes the navigation visual aid supporting a watch officer's situation awareness and analyzes its performance test result. Developing the equipment made from composite video sensor which transfer video signal, ranger laser measurement model which search out distance, Pan/ Tilt, center control device. The developed equipment with Pan/Tilt was made from high performance video sensor and ranger laser measurement. To make a real ship test, we carried on setting the developed equipment on ship, observed a danger factor and analyzed a image, and from that we can evaluate marine environment awareness. Through this result, the developed equipment can show effective ability of the awareness of the clearer check and resolution situation when compare with the binocular.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.2
• /
• pp.79-86
• /
• 2024

In this paper, a novel visual tracking method which can utilize the context of object regions is presented. Conventional methods have the inherent problem of treating all candidate regions independently, where the tracker could not successfully discriminate regions with similar appearances. This was due to lack of contextual modeling in a given scene, where all candidate object regions should be taken into consideration when choosing a single region. The goal of the proposed method is to encourage feature exchange between candidate regions to improve the discriminability between similar regions. It improves upon conventional methods that only consider a single region, and is implemented by employing the MLP-Mixer model for enhanced feature exchange between regions. By implementing channel-wise, inter-region interaction operation between candidate features, contextual information of regions can be embedded into the individual feature representations. To evaluate the performance of the proposed tracker, the large-scale LaSOT dataset is used, and the experimental results show a competitive AUC performance of 0.560 while running at a real-time speed of 65 fps.

• Journal of Biomedical Engineering Research
• /
• v.45 no.2
• /
• pp.101-107
• /
• 2024

Steady-state visual evoked potential-based brain-computer interface (SSVEP-BCI) is one of the promising systems that can serve as an alternative input device due to its stable and fast performance. However, one of the major bottlenecks is that some individuals exhibit no or very low SSVEP responses to flickering stimulation, known as SSVEP illiteracy, resulting in low performance on SSVEP-BCIs. However, a lengthy duration is required to enhance multiple SSVEP responses using traditional single-frequency transcranial alternating current stimulation (tACS). This research proposes a novel approach using dual-frequency tACS (df-tACS) to potentially enhance SSVEP by targeting the two frequencies with the lowest signal-to-noise ratio (SNR) for each participant. Seven participants (five males, average age: 24.42) were exposed to flickering checkerboard stimuli at six frequencies to determine the weakest SNR frequencies. These frequencies were then simultaneously stimulated using df-tACS for 20 minutes, and the experiment was repeated to evaluate changes in SSVEP responses. The results showed that df-tACS effectively enhances the SNR at each targeted frequency, suggesting it can selectively improve target frequency responses. The study supports df-tACS as a more efficient solution for SSVEP illiteracy, proposing further exploration into multi-frequency tACS that could stimulate more than two frequencies, thereby expanding the potential of SSVEP-BCIs.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.12
• /
• pp.485-496
• /
• 2018

So far, the deep learning, a field of artificial intelligence, has achieved remarkable results in solving problems from unstructured data. However, it is difficult to comprehensively judge situations like humans, and did not reach the level of intelligence that deduced their relations and predicted the next situation. Recently, deep neural networks show that artificial intelligence can possess powerful relational reasoning that is core intellectual ability of human being. In this paper, to analyze and observe the performance of Relation Networks (RN) among the neural networks for relational reasoning, two types of RN-based deep neural network models were constructed and compared with the baseline model. One is a visual question answering RN model using Sort-of-CLEVR and the other is a text-based question answering RN model using bAbI task. In order to maximize the performance of the RN-based model, various performance improvement experiments such as hyper parameters tuning have been proposed and performed. The effectiveness of the proposed performance improvement methods has been verified by applying to the visual QA RN model and the text-based QA RN model, and the new domain model using the dialogue-based LL dataset. As a result of the various experiments, it is found that the initial learning rate is a key factor in determining the performance of the model in both types of RN models. We have observed that the optimal initial learning rate setting found by the proposed random search method can improve the performance of the model up to 99.8%.

• Journal of Korean Ophthalmic Optics Society
• /
• v.14 no.3
• /
• pp.43-49
• /
• 2009

Purpose: The purpose of this study was to obtain the fundamental data of dynamic visual acuity and dynamic stereoacuity. Methods: The subjects were 20 athletes (high school baseball player) and 40 nonathletes (20 male, 20 female). We assessed static visual acuity, dynamic visual acuity, static stereoacuity and dynamic stereoacuity using rotating mirror projection system and computer program. Results: Three groups had similar static visual acuity and static stereoacuity. On the other hand, the dynamic visual acuity and dynamic stereoacuity showed statistically significant differency. The mean dynamic visual acuity for athletes was 174.80${\pm}$28.70 deg/sec, 137.10${\pm}$16.54 deg/sec for male nonathletes and 111.59${\pm}$15.40 deg/sec for female nonathletes. The mean dynamic stereoacuity for athlets was 234.55${\pm}$19.64, 249.05${\pm}$8.86 for male nonathletes and 247.10${\pm}$14.89 for female nonathletes. The group of athletes had better dynamic visual acuity and dynamic stereoacuity. Conclusions: If the result of this study apply to sports, it will be very useful to improve sports performance.