• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.12 no.6
• /
• pp.591-596
• /
• 2019

In this paper, we implemented a deep learning operation structure with less influence of local PC performance. In general, the deep learning model has a large amount of computation and is heavily influenced by the performance of the processing PC. In this paper, we implemented deep learning operation using AWS and streaming server to reduce this limitation. First, deep learning operations were performed on AWS so that deep learning operation would work even if the performance of the local PC decreased. However, with AWS, the output is less real-time relative to the input when computed. Second, we use streaming server to increase the real-time of deep learning model. If the streaming server is not used, the real-time performance is poor because the images must be processed one by one or by stacking the images. We used the YOLO v3 model as a deep learning model for performance comparison experiments, and compared the performance of local PCs with instances of AWS and GTX1080, a high-performance GPU. The simulation results show that the test time per image is 0.023444 seconds when using the p3 instance of AWS, which is similar to the test time per image of 0.027099 seconds on a local PC with the high-performance GPU GTX1080.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.1
• /
• pp.305-318
• /
• 2022

Abusive behaviors have become a common issue in many online social media platforms. Profanity is common form of abusive behavior in online. Social media platforms operate the filtering system using popular profanity words lists, but this method has drawbacks that it can be bypassed using an altered form and it can detect normal sentences as profanity. Especially in Korean language, the syllable is composed of graphemes and words are composed of multiple syllables, it can be decomposed into graphemes without impairing the transmission of meaning, and the form of a profane word can be seen as a different meaning in a sentence. This work focuses on the problem of filtering system mis-detecting normal phrases with profane phrases. For that, we proposed the deep learning-based framework including grapheme and syllable separation-based word embedding and appropriate CNN structure. The proposed model was evaluated on the chatting contents from the one of the famous online games in South Korea and generated 90.4% accuracy.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.644-645
• /
• 2021

In this paper, we propose a hardware structure of a watermarking processor based on deep learning technology to protect the intellectual property rights of ultra-high resolution digital images and videos. We propose an optimization methodology to implement a deep learning-based watermarking algorithm in hardware.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.458-461
• /
• 2021

유동인구 예측은 상권의 특성에 따른 점포의 입지 선정 및 고객 맞춤형 마케팅 등 민간 분야에서부터 교통망 등 사회 간접 자본 설계를 위한 공공 분야에 이르기까지 다양한 목적으로 연구되어 왔으며, 최근에는 Covid-19 의 확산에 따라 그 중요도가 더욱 높아지고 있다. 보다 정교한 예측을 위해서는 전체적인 유동 인구 뿐만 아니라 특성 별로 세분화된 하위 그룹에 대해서도 정확한 예측이 요구되나, 기존의 예측 모델들은 이러한 데이터의 계층 구조를 고려하지 않았다. 본 연구에서는 세분화된 하위 그룹 별 유동인구의 예측 정확도를 높이기 위해 전체 유동인구의 패턴을 동시에 활용하는 Global-Local 구조 기반의 Deep Learning 유동인구 분석 모델을 제안한다. 실험 결과 단일 시계열 데이터만을 사용하는 경우 대비 5.4%~52.6%의 예측 오류 감소 효과가 있음을 확인하였다.

• Journal of Korea Multimedia Society
• /
• v.19 no.5
• /
• pp.848-855
• /
• 2016

Face recognition has received a great deal of attention for the wide range of applications in real-world scenario. In this scenario, mismatches (so called heterogeneity) in terms of resolution and illumination between gallery and test face images are inevitable due to the different capturing conditions. In order to deal with the mismatch problem, we propose a local feature learning method using deep canonical correlation analysis (DCCA) for heterogeneous face recognition. By the DCCA, we can effectively reduce the mismatch between the gallery and the test face images. Furthermore, the proposed local feature learned by the DCCA is able to enhance the discriminative power by using facial local structure information. Through the experiments on two different scenarios (i.e., matching near-infrared to visible face images and matching low-resolution to high-resolution face images), we could validate the effectiveness of the proposed method in terms of recognition accuracy using publicly available databases.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.17 no.4
• /
• pp.53-62
• /
• 2021

Deep learning models such as convolutional neural networks and recurrent neual networks process a huge amounts of data, so they require a lot of storage and consume a lot of time and power due to memory access. Recently, research is being conducted to reduce memory usage and access by compressing data using the feature that many of deep learning data are highly sparse and localized. In this paper, we propose a compression-decompression method of storing only the non-zero data and the location information of the non-zero data excluding zero data. In order to make the location information of non-zero data, the matrix data is divided into sections uniformly. And whether there is non-zero data in the corresponding section is indicated. In this case, section division is not executed only once, but repeatedly executed, and location information is stored in each step. Therefore, it can be properly compressed according to the ratio and distribution of zero data. In addition, we propose a hardware structure that enables compression and decompression without complex operations. It was designed and verified with Verilog, and it was confirmed that it can be used in hardware deep learning accelerators.

• Smart Structures and Systems
• /
• v.30 no.3
• /
• pp.287-301
• /
• 2022

The machine-vision based structural displacement measurement methods are widely used due to its flexible deployment and non-contact measurement characteristics. The accuracy of vision measurement is directly related to the image resolution. In the field of computer vision, super-resolution reconstruction is an emerging method to improve image resolution. Particularly, the deep-learning based image super-resolution methods have shown great potential for improving image resolution and thus the machine-vision based measurement. In this article, we firstly review the latest progress of several deep learning based super-resolution models, together with the public benchmark datasets and the performance evaluation index. Secondly, we construct a binocular visual measurement platform to measure the distances of the adjacent corners on a chessboard that is universally used as a target when measuring the structure displacement via machine-vision based approaches. And then, several typical deep learning based super resolution algorithms are employed to improve the visual measurement performance. Experimental results show that super-resolution reconstruction technology can improve the accuracy of distance measurement of adjacent corners. According to the experimental results, one can find that the measurement accuracy improvement of the super resolution algorithms is not consistent with the existing quantitative performance evaluation index. Lastly, the current challenges and future trends of super resolution algorithms for visual measurement applications are pointed out.

• International conference on construction engineering and project management
• /
• 2022.06a
• /
• pp.814-822
• /
• 2022

The deployment of sensors for Structural Health Monitoring requires a complicated network arrangement, ground truthing, and calibration for validating sensor performance periodically. Any conventional sensor on a structural element is also subjected to static and dynamic vertical loadings in conjunction with other environmental factors, such as brightness, noise, temperature, and humidity. A structural model with strain gauges was built and tested to get realistic sensory information. This paper investigates different deep learning architectures and algorithms, including unsupervised, autoencoder, and supervised methods, to benchmark blind drift calibration methods using deep learning. It involves a fully connected neural network (FCNN), a long short-term memory (LSTM), and a gated recurrent unit (GRU) to address the blind drift calibration problem (i.e., performing calibrations of installed sensors when ground truth is not available). The results show that the supervised methods perform much better than unsupervised methods, such as an autoencoder, when ground truths are available. Furthermore, taking advantage of time-series information, the GRU model generates the most precise predictions to remove the drift overall.

• Geomechanics and Engineering
• /
• v.33 no.2
• /
• pp.203-209
• /
• 2023

One major advantage of ground penetrating radar (GPR) over other field test methods is its ability to obtain subsurface images of roads in an efficient and non-intrusive manner. Not only can the strata of pavement structure be retrieved from the GPR scan images, but also various irregularities, such as cracks and internal cavities. This article introduces a deep learning-based approach, focusing on detecting subsurface cracks by recognizing their distinctive hyperbolic signatures in the GPR scan images. Given the limited road sections that contain target features, two data augmentation methods, i.e., feature insertion and generation, are implemented, resulting in 9,174 GPR scan images. One of the most popular real-time object detection models, You Only Learn One Representation (YOLOR), is trained for detecting the target features for two types of subsurface cracks: bottom cracks and full cracks from the GPR scan images. The former represents partial cracks initiated from the bottom of the asphalt layer or base layers, while the latter includes extended cracks that penetrate these layers. Our experiments show the test average precisions of 0.769, 0.803 and 0.735 for all cracks, bottom cracks, and full cracks, respectively. This demonstrates the practicality of deep learning-based methods in detecting subsurface cracks from GPR scan images.

• Journal of IKEEE
• /
• v.25 no.3
• /
• pp.493-500
• /
• 2021

In this paper, we propose the development of deep learning structure to improve quality of polygonal containers. The deep learning structure consists of a convolution layer, a bottleneck layer, a fully connect layer, and a softmax layer. The convolution layer is a layer that obtains a feature image by performing a convolution 3x3 operation on the input image or the feature image of the previous layer with several feature filters. The bottleneck layer selects only the optimal features among the features on the feature image extracted through the convolution layer, reduces the channel to a convolution 1x1 ReLU, and performs a convolution 3x3 ReLU. The global average pooling operation performed after going through the bottleneck layer reduces the size of the feature image by selecting only the optimal features among the features of the feature image extracted through the convolution layer. The fully connect layer outputs the output data through 6 fully connect layers. The softmax layer multiplies and multiplies the value between the value of the input layer node and the target node to be calculated, and converts it into a value between 0 and 1 through an activation function. After the learning is completed, the recognition process classifies non-circular glass bottles by performing image acquisition using a camera, measuring position detection, and non-circular glass bottle classification using deep learning as in the learning process. In order to evaluate the performance of the deep learning structure to improve quality of polygonal containers, as a result of an experiment at an authorized testing institute, it was calculated to be at the same level as the world's highest level with 99% good/defective discrimination accuracy. Inspection time averaged 1.7 seconds, which was calculated within the operating time standards of production processes using non-circular machine vision systems. Therefore, the effectiveness of the performance of the deep learning structure to improve quality of polygonal containers proposed in this paper was proven.