• Journal of IKEEE
• /
• v.26 no.3
• /
• pp.364-372
• /
• 2022

This paper proposes the design and implementation results for human and object classification systems utilizing frequency modulated continuous wave (FMCW) radar sensor. Such a system requires the process of radar sensor signal processing for multi-target detection and the process of deep learning for the classification of human and object. Since deep learning requires such a great amount of computation and data processing, the lightweight process is utmost essential. Therefore, binary neural network (BNN) structure was adopted, operating convolution neural network (CNN) computation in a binary condition. In addition, for the real-time operation, a hardware accelerator was implemented and verified via FPGA platform. Based on performance evaluation and verified results, it is confirmed that the accuracy for multi-target classification of 90.5%, reduced memory usage by 96.87% compared to CNN and the run time of 5ms are achieved.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.10a
• /
• pp.808-810
• /
• 2018

Pedestrian tracking is a particular problem of object tracking, and an important component in various vision-based applications, such as autonomous cars or surveillance systems. After several years of development, pedestrian tracking in videos is still a challenging problem because of various visual properties of objects and surrounding environment. In this research, we propose a tracking-by-detection system for pedestrian tracking, which incorporates Convolutional Neural Network (CNN) and color information. Pedestrians in video frames are localized by a CNN, then detected pedestrians are assigned to their corresponding tracklets based on similarities in color distributions. The experimental results show that our system was able to overcome various difficulties to produce highly accurate tracking results.

• Journal of the Semiconductor & Display Technology
• /
• v.19 no.4
• /
• pp.71-76
• /
• 2020

In this paper, we propose a super-resolution method that reconstructs compressed low-resolution images into high-resolution images. We propose a CNN model with a small number of parameters, and even if quantization is applied to the proposed model, super-resolution can be implemented without deteriorating the image quality. To further improve the quality of the compressed low-resolution image, a new degradation model was proposed instead of the existing bicubic degradation model. The proposed degradation model is used only in the training process and can be applied by changing only the parameter values to the original CNN model. In the super-resolution image applying the proposed degradation model, visual artifacts caused by image compression were effectively removed. As a result, our proposed method generates higher PSNR values at compressed images and shows better visual quality, compared to conventional CNN-based SR methods.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2023.11a
• /
• pp.261-262
• /
• 2023

Drone is a suitable equipment for capturing images of cracks at construction sites based on its efficient mobility and high-resolution image acquisition capabilities. In this study, drone was used to acquire indoor construction sites framework images and deep learning technology was applied to detect cracks and measure width, and size. Finally, the usability of the process was verified based on the indoor crack monitoring system.

• Journal of KIISE
• /
• v.44 no.7
• /
• pp.680-685
• /
• 2017

The emergence of online media and their data has enabled data-driven methods to solve challenging and complex tasks such as rumor classification problems. Recently, deep learning based models have been shown as one of the fastest and the most accurate algorithms to solve such problems. These new models, however, either rely on complete data or several days-worth of data, limiting their applicability in real time. In this study, we go beyond this limit and test the possibility of super early rumor detection via recurrent neural networks (RNNs). Our model takes in social media streams as time series input, along with basic meta-information about the rumongers including the follower count and the psycholinguistic traits of rumor content itself. Based on analyzing millions of social media posts on 498 real rumors and 494 non-rumor events, our RNN-based model detected rumors with only 30 initial posts (i.e., within a few hours of rumor circulation) with remarkable F1 score of 0.74. This finding widens the scope of new possibilities for building a fast and efficient rumor detection system.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.1
• /
• pp.29-35
• /
• 2019

As the fourth industrial revolution is emerging, many companies are increasingly interested in smart factories and the importance of sensors is being emphasized. In the case that sensors for collecting sensing data fail, the plant could not be optimized and further it could not be operated properly, which may incur a financial loss. For this purpose, it is necessary to diagnose the status of sensors to prevent sensor' fault. In the paper, we propose a scheme to diagnose digital-sensor' fault by analyzing the rising time and falling time of digital sensors through the LSTM(Long Short Term Memory) of Deep Learning RNN algorithm. Experimental results of the proposed scheme are compared with those of rule-based fault diagnosis algorithm in terms of AUC(Area Under the Curve) of accuracy and ROC(Receiver Operating Characteristic) curve. Experimental results show that the proposed system has better and more stable performance than the rule-based fault diagnosis algorithm.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.20 no.1
• /
• pp.70-85
• /
• 2021

Deep learning methodology, which has been actively studied in recent years, has improved the performance of artificial intelligence. Accordingly, systems utilizing deep learning have been proposed in various industries. In traffic systems, spatio-temporal graph modeling using GNN was found to be effective in predicting traffic speed. Still, it has a disadvantage that the model is trained inefficiently due to the memory bottleneck. Therefore, in this study, the road network is clustered through the graph clustering algorithm to reduce memory bottlenecks and simultaneously achieve superior performance. In order to verify the proposed method, the similarity of road speed distribution was measured using Jensen-Shannon divergence based on the analysis result of Incheon UTIC data. Then, the road network was clustered by spectrum clustering based on the measured similarity. As a result of the experiments, it was found that when the road network was divided into seven networks, the memory bottleneck was alleviated while recording the best performance compared to the baselines with MAE of 5.52km/h.

• Smart Structures and Systems
• /
• v.30 no.6
• /
• pp.613-626
• /
• 2022

Guaranteeing the quality and integrity of structural health monitoring (SHM) data is very important for an effective assessment of structural condition. However, sensory system may malfunction due to sensor fault or harsh operational environment, resulting in multiple types of data anomaly existing in the measured data. Efficiently and automatically identifying anomalies from the vast amounts of measured data is significant for assessing the structural conditions and early warning for structural failure in SHM. The major challenges of current automated data anomaly detection methods are the imbalance of dataset categories. In terms of the feature of actual anomalous data, this paper proposes a data anomaly detection method based on data-level and deep learning technique for SHM of civil engineering structures. The proposed method consists of a data balancing phase to prepare a comprehensive training dataset based on data-level technique, and an anomaly detection phase based on a sophisticatedly designed network. The advanced densely connected convolutional network (DenseNet) and Transformer encoder are embedded in the specific network to facilitate extraction of both detail and global features of response data, and to establish the mapping between the highest level of abstractive features and data anomaly class. Numerical studies on a steel frame model are conducted to evaluate the performance and noise immunity of using the proposed network for data anomaly detection. The applicability of the proposed method for data anomaly classification is validated with the measured data of a practical supertall structure. The proposed method presents a remarkable performance on data anomaly detection, which reaches a 95.7% overall accuracy with practical engineering structural monitoring data, which demonstrates the effectiveness of data balancing and the robust classification capability of the proposed network.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.46 no.4
• /
• pp.152-159
• /
• 2023

This study endeavors to enrich investment prospects in cryptocurrency by establishing a rationale for investment decisions. The primary objective involves evaluating the predictability of four prominent cryptocurrencies - Bitcoin, Ethereum, Litecoin, and EOS - and scrutinizing the efficacy of trading strategies developed based on the prediction model. To identify the most effective prediction model for each cryptocurrency annually, we employed three methodologies - AutoRegressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM), and Prophet - representing traditional statistics and artificial intelligence. These methods were applied across diverse periods and time intervals. The result suggested that Prophet trained on the previous 28 days' price history at 15-minute intervals generally yielded the highest performance. The results were validated through a random selection of 100 days (20 target dates per year) spanning from January 1st, 2018, to December 31st, 2022. The trading strategies were formulated based on the optimal-performing prediction model, grounded in the simple principle of assigning greater weight to more predictable assets. When the forecasting model indicates an upward trend, it is recommended to acquire the cryptocurrency with the investment amount determined by its performance. Experimental results consistently demonstrated that the proposed trading strategy yields higher returns compared to an equal portfolio employing a buy-and-hold strategy. The cryptocurrency trading model introduced in this paper carries two significant implications. Firstly, it facilitates the evolution of cryptocurrencies from speculative assets to investment instruments. Secondly, it plays a crucial role in advancing deep learning-based investment strategies by providing sound evidence for portfolio allocation. This addresses the black box issue, a notable weakness in deep learning, offering increased transparency to the model.

• Korean Journal of Radiology
• /
• v.23 no.1
• /
• pp.139-149
• /
• 2022

Objective: To compare the effects of bone suppression imaging using deep learning (BSp-DL) based on a generative adversarial network (GAN) and bone subtraction imaging using a dual energy technique (BSt-DE) on radiologists' performance for pulmonary nodule detection on chest radiographs (CXRs). Materials and Methods: A total of 111 adults, including 49 patients with 83 pulmonary nodules, who underwent both CXR using the dual energy technique and chest CT, were enrolled. Using CT as a reference, two independent radiologists evaluated CXR images for the presence or absence of pulmonary nodules in three reading sessions (standard CXR, BSt-DE CXR, and BSp-DL CXR). Person-wise and nodule-wise performances were assessed using receiver-operating characteristic (ROC) and alternative free-response ROC (AFROC) curve analyses, respectively. Subgroup analyses based on nodule size, location, and the presence of overlapping bones were performed. Results: BSt-DE with an area under the AFROC curve (AUAFROC) of 0.996 and 0.976 for readers 1 and 2, respectively, and BSp-DL with AUAFROC of 0.981 and 0.958, respectively, showed better nodule-wise performance than standard CXR (AUAFROC of 0.907 and 0.808, respectively; p ≤ 0.005). In the person-wise analysis, BSp-DL with an area under the ROC curve (AUROC) of 0.984 and 0.931 for readers 1 and 2, respectively, showed better performance than standard CXR (AUROC of 0.915 and 0.798, respectively; p ≤ 0.011) and comparable performance to BSt-DE (AUROC of 0.988 and 0.974; p ≥ 0.064). BSt-DE and BSp-DL were superior to standard CXR for detecting nodules overlapping with bones (p < 0.017) or in the upper/middle lung zone (p < 0.017). BSt-DE was superior (p < 0.017) to BSp-DL in detecting peripheral and sub-centimeter nodules. Conclusion: BSp-DL (GAN-based bone suppression) showed comparable performance to BSt-DE and can improve radiologists' performance in detecting pulmonary nodules on CXRs. Nevertheless, for better delineation of small and peripheral nodules, further technical improvements are required.