• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.2
• /
• pp.63-70
• /
• 2022

Although 2D Object detection has been largely improved in the past years with the advance of deep learning methods and the use of large labeled image datasets, 3D object detection from 2D imagery is a challenging problem in a variety of applications such as robotics, due to the lack of data and diversity of appearances and shapes of objects within a category. Google has just announced the launch of Objectron that has a novel data pipeline using mobile augmented reality session data. However, it also is corresponding to 2D-driven 3D object detection technique. This study explores more mature 2D object detection method, and applies its 2D projection to Objectron 3D lifting system. Most object detection methods use bounding boxes to encode and represent the object shape and location. In this work, we explore a stochastic representation of object regions using Gaussian distributions. We also present a similarity measure for the Gaussian distributions based on the Hellinger Distance, which can be viewed as a stochastic Intersection-over-Union. Our experimental results show that the proposed Gaussian representations are closer to annotated segmentation masks in available datasets. Thus, less accuracy problem that is one of several limitations of Objectron can be relaxed.

• Journal of Broadcast Engineering
• /
• v.27 no.1
• /
• pp.44-55
• /
• 2022

In order to automatically provide information on fruits in agricultural related broadcasting contents, instance image segmentation of target fruits is required. In addition, the information on the 3D pose of the corresponding fruit may be meaningfully used. This paper represents research that provides information about tomatoes in video content. A large amount of data is required to learn the instance segmentation, but it is difficult to obtain sufficient training data. Therefore, the training data is generated through a data augmentation technique based on a small amount of real images. Compared to the result using only the real images, it is shown that the detection performance is improved as a result of learning through the synthesized image created by separating the foreground and background. As a result of learning augmented images using images created using conventional image pre-processing techniques, it was shown that higher performance was obtained than synthetic images in which foreground and background were separated. To estimate the pose from the result of object detection, a point cloud was obtained using an RGB-D camera. Then, cylinder fitting based on least square minimization is performed, and the tomato pose is estimated through the axial direction of the cylinder. We show that the results of detection, instance image segmentation, and cylinder fitting of a target object effectively through various experiments.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.8 no.9
• /
• pp.265-273
• /
• 2018

In this paper, we propose an acoustic event detection method in tunnels using non-negative tensor factorization (NTF) and hidden Markov model (HMM) applied to multi-channel audio signals. Incidents in tunnel are inherent to the system and occur unavoidably with known probability. Incidents can easily happen minor accidents and extend right through to major disaster. Most incident detection systems deploy visual incident detection (VID) systems that often cause false alarms due to various constraints such as night obstacles and a limit of viewing angle. To this end, the proposed method first tries to separate and detect every acoustic event, which is assumed to be an in-tunnel incident, from noisy acoustic signals by using an NTF technique. Then, maximum likelihood estimation using Gaussian mixture model (GMM)-HMMs is carried out to verify whether or not each detected event is an actual incident. Performance evaluation shows that the proposed method operates in real time and achieves high detection accuracy under simulated tunnel conditions.

• Journal of Industrial Convergence
• /
• v.21 no.9
• /
• pp.41-48
• /
• 2023

Recently, there has been a sharp increase in the damages caused by ransomware across various sectors of society, including individuals, businesses, and nations. Ransomware is a malicious software that infiltrates user computer systems, encrypts important files, and demands a ransom in exchange for restoring access to the files. Due to its diverse and sophisticated attack techniques, ransomware is more challenging to detect than other types of malware, and its impact is significant. Therefore, there is a critical need for accurate detection and mitigation methods. To achieve precise ransomware detection, an inference engine of a detection system must possess knowledge of ransomware features. In this paper, we propose a model to extract and classify the characteristics of ransomware for accurate detection of ransomware, calculate the similarity of the extracted characteristics, reduce the dimension of the characteristics, group the reduced characteristics, and classify the characteristics of ransomware into attack tools, inflow paths, installation files, command and control, executable files, acquisition rights, circumvention techniques, collected information, leakage techniques, and state changes of the target system. The classified characteristics were applied to the existing ransomware to prove the validity of the classification, and later, if the inference engine learned using this classification technique is installed in the detection system, most of the newly emerging and variant ransomware can be detected.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.7
• /
• pp.41-51
• /
• 2024

In this paper, we explore the enhancement of target detection accuracy in the guided weapon using deep learning object detection on infrared (IR) images. Due to the characteristics of IR images being influenced by factors such as time and temperature, it's crucial to ensure a consistent representation of object features in various environments when training the model. A simple way to address this is by emphasizing the features of target objects and reducing noise within the infrared images through appropriate pre-processing techniques. However, in previous studies, there has not been sufficient discussion on pre-processing methods in learning deep learning models based on infrared images. In this paper, we aim to investigate the impact of image pre-processing techniques on infrared image-based training for object detection. To achieve this, we analyze the pre-processing results on infrared images that utilized global or local information from the video and the image. In addition, in order to confirm the impact of images converted by each pre-processing technique on object detector training, we learn the YOLOX target detector for images processed by various pre-processing methods and analyze them. In particular, the results of the experiments using the CLAHE (Contrast Limited Adaptive Histogram Equalization) shows the highest detection accuracy with a mean average precision (mAP) of 81.9%.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.3
• /
• pp.148-155
• /
• 2024

Despite the increasing importance of irony and sarcasm detection in the field of natural language processing, research on the Korean language is relatively scarce compared to other languages. This study aims to experiment with various models for irony detection in Korean text. The study conducted irony detection experiments using KoBERT, a BERT-based model, and ChatGPT. For KoBERT, two methods of additional training on sentiment data were applied (Transfer Learning and MultiTask Learning). Additionally, for ChatGPT, the Few-Shot Learning technique was applied by increasing the number of example sentences entered as prompts. The results of the experiments showed that the Transfer Learning and MultiTask Learning models, which were trained with additional sentiment data, outperformed the baseline model without additional sentiment data. On the other hand, ChatGPT exhibited significantly lower performance compared to KoBERT, and increasing the number of example sentences did not lead to a noticeable improvement in performance. In conclusion, this study suggests that a model based on KoBERT is more suitable for irony detection than ChatGPT, and it highlights the potential contribution of additional training on sentiment data to improve irony detection performance.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.44 no.2
• /
• pp.15-21
• /
• 2007

In this paper, we propose a technique of a practical symbol acquisition and tracking using a low complex ADC and simple digital circuits for noncoherent asynchronous impulse-radio-based Ultra Wideband (IR-UWB) receiver based on energy detection. Compared to previous approaches of detecting an exact acquisition time that require much hardware resource, the proposed technique is to detect the target symbol by finding the symbol acquisition interval per symbol with a target symbo, thus the complexity of the complete signal processing and power consumption by ADC are reduced. To do this, we define the bit decision window (BDW) and analyze the relation between SNR, hardware resource, size of BDW and BER(Bit Error Rate). Using the results, the optimum BDW size for the minimum BER with limited hardware resource is selected. The proposed synchronization technique is verified with an aid of a simulator programmed by considering practical impulse channels.

• Smart Structures and Systems
• /
• v.14 no.5
• /
• pp.811-830
• /
• 2014

A numerical study has been carried out to simulate an innovative monitoring procedure to detect and localize damage in reinforced concrete beams retrofitted with carbon fiber reinforced polymer (CFRP) unidirectional laminates. The main novelty of the present simulation is its ability to conduct the electromechanical admittance monitoring technique by considerably compressing the amount of data required for damage detection and localization. A FEM simulation of electromechanical admittance-based sensing technique was employed by applying lead zirconate titanate (PZT) transducers to acquire impedance spectrum signatures. Response surface methodology (RSM) is finally adopted as a tool for solving inverse problems to estimate the location and size of damaged areas from the relationship between damage and electromechanical admittance changes computed at PZT transducer surfaces. This statistical metamodel technique allows polynomial models to be produced without requiring complicated modeling or numerous data sets after the generation of damage, leading to considerably lower cost of creating diagnostic database. Finally, a numerical example is carried out regarding a steel-reinforced concrete (RC) beam model monotonically loaded up to its failure which is also retrofitted by a CFRP laminate to verify the validity of the present metamodeling monitoring technique. The load-carrying capacity of concrete is predicted in the present paper by utilizing an Ottosen-type failure surface in order to better take into account the passive confinement behavior of retrofitted concrete material under the application of FRP laminate.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.53 no.11
• /
• pp.768-775
• /
• 2004

The SSA technique in the digital circuit test is required to be repeated the input pattern stream to n bits output nodes n times in case of using a multiplexor. Because the method adopting a parallel/serial bit convertor to remove this inefficiency has disadvantage of requiring the test time n times for a pattern, the test strategy is required, which can enhance the test productivity by reducing the test time based on simplified fault detection mechanism. Accordingly, this paper proposes a test strategy which enhances the test productivity and efficiency by appling PAS (Parallel Signature Analysis) technique to those after analyzing the structure and characteristics of the digital devices including TTL and CMOS family ICs as well as ROM and RAM. The PSA technique identifies the faults by comparing the reminder from good device with reminder from the tested device. At this time, the reminder is obtained by enforcing the data stream obtained from output pins of the tested device on the LFSR(Linear Feedback Shift Resister) representing the characteristic equation. Also, the method to obtain the optimal signature analyzer is explained by furnishing the short bit input streams to the long bit input streams to the LFSR having 8, 12, 16, 20bit input/output pins and by analyzing the occurring probability of error which is impossible to detect. Finally, the effectiveness of the proposed test strategy is verified by simulating the stuck at 1 errors or stuck at 0 errors for several devices on typical 8051 digital board.

• Nuclear Engineering and Technology
• /
• v.52 no.3
• /
• pp.621-625
• /
• 2020

Cast austenitic stainless steel (CASS) is used for fabricating different components of the primary reactor coolant system of pressurized water reactors. However, the thermal embrittlement of CASS resulting from long-term operation causes structural safety problems. Ultrasonic testing for flaw detection has been used to assess the thermal embrittlement of CASS; however, the high scattering and attenuation of the ultrasonic wave propagating through CASS make it difficult to accurately quantify the flaw size. In this paper, we present a different approach for evaluating the thermal embrittlement of CASS by assessing changes in the material properties of CASS using a nonlinear ultrasonic technique, which is a potential nondestructive method. For the evaluation, we prepared CF8M specimens that were thermally aged under four different heating conditions. Nonlinear ultrasonic measurements were performed using a contact piezoelectric method to obtain the relative ultrasonic nonlinearity parameter, and a mini-sized tensile test was performed to investigate the correlation of the parameter with material properties. Experimental results showed that the ultrasonic nonlinearity parameter had a correlation with tensile properties such as the tensile strength and elongation. Consequently, we could confirm the applicability of the nonlinear ultrasonic technique to the evaluation of the thermal embrittlement of CASS.