• Journal of Platform Technology
• /
• v.12 no.1
• /
• pp.128-140
• /
• 2024

'Deepfake' refers to a video synthesis technique that utilizes various artificial intelligence technologies to create highly realistic fake content, causing serious confusion to individuals and society by being used for generating fake news, fraud, malicious impersonation, and more. To address this issue, there is a need for methods to detect malicious images generated by deepfake accurately. In this paper, we extract and analyze saliency features from deepfake and real images, and detect candidate synthesis regions on the images, and finally construct an automatic deepfake detection model by focusing on the extracted features. The proposed saliency feature-based model can be universally applied in situations where deepfake detection is required, such as synthesized images and videos. To demonstrate the performance of our approach, we conducted several experiments that have shown the effectiveness of the deepfake detection task.

• Korean Journal of Remote Sensing
• /
• v.36 no.5_4
• /
• pp.1245-1254
• /
• 2020

In this study, the limitations of the existing water level detection algorithm using CCTV images were recognized and the water level detection algorithm was improved by applying the Region growing technique. It applied three techniques (Horizontal projection profile, Texture analysis, and Optical flow) to estimate the water area, and the results were analyzed in a comprehensive analysis to select the initial water area. The water level was then continuously detected by the Region growing technique, referring to the initial water area. As a result, it was possible to confirm that the exact level of water was detected without being affected by environmental factors compared to the existing level detection algorithm, which had frequent mis-detection phenomena depending on the surrounding environmental factors. In addition, the water level was detected in the video showing flooded roads in urban areas, not in the video of the river. These results are believed to be able to supplement the difficulty of monitoring at all times with limited manpower by automatically detecting the level of water through numerous CCTV footage installed throughout the country, and to contribute to laying the foundation for preventing disasters caused by torrential rains and typhoons in advance.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.28 no.6
• /
• pp.1044-1053
• /
• 2022

Time domain reflectometry (TDR) is a diagnostic technique to evaluate the physical integrity of cable and finds application in leak detection and localization of piping system. In this study, a cable-shaped leak detection sensor was proposed using the TDR technique for monitoring leakage detection of ship's engine room seawater piping system. The cable sensor was developed using a twisted pair arrangement and wound by an absorbent material. The availability and performance of the sensor for leak detection and localization were evaluated on a lab-scale pipeline set up. The developed sensor was installed onto the pipes and flanges of the lab-scale set up and various TDR waveforms were acquired and analyzed according to the dif erent variables including the number of twists and sheath thickness. The result indicated that the twisted cable sensor was able to produce clear and smooth signal as compared to the TDR sensor with a parallel arrangement. The optimal number of twist was determined to be above 10 per the unit length. The optimal diameter of sheath thickness that results in the desired sensitivity was determined to be ranging from 80% up to 120% of the diameter of the conductor. The linear regression analysis for estimation of leak localization was carried out to estimate the location of the leakage, and the result was a determination coefficient of 0.9998, indicating a positive relationship with the actual leakage point. The proposed TDR based leak detection method appears to be an effective method for monitoring leakage of ship's seawater piping system.

• Journal of Korean Society of Steel Construction
• /
• v.9 no.3 s.32
• /
• pp.287-300
• /
• 1997

In this study, an effective application method of AE technique for the detection of fatigue crack in multi-girder steel bridges has been proposed. The applicability has been examined through the laboratory works with bridge model. The proposed analytical method which evaluates the remaining fatigue lives of structural members may improve the rational determination of the priority of inspection for structural members assuming to have fatigue cracks. Laboratory tests for the application of AE technique to steel girder bridges show that the frequency bands of traffic noise are in the range between 10 show that the frequency bands of traffic noise are in the range between 100~200 kHz and the AE signal raised from fatigue cracks is concentrated around 400~500 kHz. Therefore. R30 sensor is proved to be the most suitable for the detection of cracks in steel girder bridges. A linear proportionality between the crack propagation and the frequency of AE signals has been obtained. In addition, an economic and effective source location method for steel girder bridges was studied through experiments.

• KIPS Transactions on Software and Data Engineering
• /
• v.4 no.10
• /
• pp.447-454
• /
• 2015

This paper proposes geographical name denoising by machine learning of event detection based on twitter. Recently, the increasing number of smart phone users are leading the growing user of SNS. Especially, the functions of short message (less than 140 words) and follow service make twitter has the power of conveying and diffusing the information more quickly. These characteristics and mobile optimised feature make twitter has fast information conveying speed, which can play a role of conveying disasters or events. Related research used the individuals of twitter user as the sensor of event detection to detect events that occur in reality. This research employed geographical name as the keyword by using the characteristic that an event occurs in a specific place. However, it ignored the denoising of relationship between geographical name and homograph, it became an important factor to lower the accuracy of event detection. In this paper, we used removing and forecasting, these two method to applied denoising technique. First after processing the filtering step by using noise related database building, we have determined the existence of geographical name by using the Naive Bayesian classification. Finally by using the experimental data, we earned the probability value of machine learning. On the basis of forecast technique which is proposed in this paper, the reliability of the need for denoising technique has turned out to be 89.6%.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.23 no.4
• /
• pp.253-263
• /
• 2021

Concrete structures are damaged by aging and external environmental factors. This type of damage is to appear in the form of cracks, to proceed in the form of spalling. Such concrete damage can act as the main cause of reducing the original design bearing capacity of the structure, and negatively affect the stability of the structure. If such damage continues, it may lead to a safety accident in the future, thus proper repair and reinforcement are required. To this end, an accurate and objective condition inspection of the structure must be performed, and for this inspection, a sensor technology capable of detecting damage area is required. For this reason, we propose a deep learning-based image processing algorithm that can detect spalling. To develop this, 298 spalling images were obtained, of which 253 images were used for training, and the remaining 45 images were used for testing. In addition, an improved loss function and data augmentation technique were applied to improve the detection performance. As a result, the detection performance of concrete spalling showed a mean intersection over union of 80.19%. In conclusion, we developed an algorithm to detect concrete spalling through a deep learning-based image processing technique, with an improved loss function and data augmentation technique. This technology is expected to be utilized for accurate inspection and diagnosis of structures in the future.

• Korean Journal of Remote Sensing
• /
• v.30 no.3
• /
• pp.351-366
• /
• 2014

The aim of this study was to develop software for beach litter detection that includes a Graphical User Interface (GUI) and uses images taken by a micro-unmanned aerial vehicle. Videos were taken over Doomo pebble beach, Sogye pebble beach, and Heungnam sand beach on the northeast coast of Geojedo (Geoje Island), Korea. Still images of actual beach litter were obtained from the videos. The image processing involved preprocessing, morphological image processing, and image recognition. Comparison with still images showing beach litter demonstrated that the software could generally detect litter larger than 50 cm in size such as Styrofoam buoys and circular fish traps (excluding small pixel-size ropes). Combining the proposed method with the conventional surveying approach is expected to enhance the accuracy of beach litter detection. The new technique will also aid in predicting the amount of beach litter generated along coastlines, which is currently difficult to monitor.

• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.852-856
• /
• 2004

An induction motor is one of the most popular electrical apparatuses owing to its simple structure and robust construction. Parameter identification of the induction motor has long been researched either for a vector control technique or fault detection. Since vector control is a well-established technique for induction motor control, this paper concentrates on successive identification of physical parameters with on-load data for the purpose of condition monitoring and/or fault detection. For extracting six physical parameters from the on-load data in the framework of the induction motor state equation, unmeasured initial state values and profiles of load torque have to be estimated as well. However, the analytic optimization methods in general fail to estimate these auxiliary but significant parameters owing to the difficulty of obtaining their gradient information. In this paper, the univariate dynamic encoding algorithm for searches (uDEAS) newly developed is applied to the identification of whole unknown parameters in the mathematical equations of an induction motor with normal operating data. Profiles of identified parameters appear to be reasonable and therefore the proposed approach is available for fault diagnosis of induction motors by monitoring physical parameters.

• Journal of Korean Society of Water and Wastewater
• /
• v.18 no.5
• /
• pp.628-637
• /
• 2004

In many drinking water treatment plants, chlorination process is one of the main techniques used for the disinfection of water. This disinfecting treatment leads to the formation of disinfection by-products (DBPs) such as haloacetonitriles (HANs), trihalomethanes (THMs), haloacetic acids (HAAs). In this study, headspace-solid phase microextraction (HS- SPME) technique was applied for the analysis of HANs in drinking water. The effects of experimental parameters such as selection of SPME fiber, the addition of salts, magnetic stirring, extraction temperature, extraction time and desorption time on the analysis were investigated. Analytical parameters such as linearity, repeatability and detection limits were also evaluated. The $50/30{\mu}m$-divinylbenzene/carboxen/polydimethylsiloxane fiber, extraction time of 30 minutes, extraction temperature of $20^{\circ}C$ and desorption time of 1 minute at $260^{\circ}C$ were the optimal experimental conditions for the analysis of HANs. The correlation coefficients ($r^2$) for HANs was 0.9979~0.9991, respectively. The relative standard deviations (%RSD) for HANs was 2.3~7.6%, respectively. Detection limits (LDs) for HANs was $0.01{\sim}0.5{\mu}g/L$, respectively.

• Smart Structures and Systems
• /
• v.22 no.5
• /
• pp.587-609
• /
• 2018

There is greater significance in identifying the incipient damages in structures at the time of their initiation as timely rectification of these minor incipient cracks can save huge maintenance cost. However, the change in the global dynamic characteristics of a structure due to these subtle damages are insignificant enough to detect using the majority of the current damage diagnostic techniques. Keeping this in view, we propose a hybrid damage diagnostic technique for detection of minor incipient damages in the structures. In the proposed automated hybrid algorithm, the raw dynamic signatures obtained from the structure are decomposed to uni-modal signals and the dynamic signature are reconstructed by identifying and combining only the uni-modal signals altered by the minor incipient damage. We use these reconstructed signals for damage diagnostics using ARMAX model. Numerical simulation studies are carried out to investigate and evaluate the proposed hybrid damage diagnostic algorithm and their capability in identifying minor/incipient damage with noisy measurements. Finally, experimental studies on a beam are also presented to compliment the numerical simulations in order to demonstrate the practical application of the proposed algorithm.