• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.1311-1314
• /
• 2022

DNN을 사용하여 객체 인식 과정에서 객체를 잘 분류하기 위해서는 시각적 설명성이 요구된다. 시각적 설명성은 object class에 대한 예측을 pixel-wise attribution으로 표현해 예측 근거를 해석하기 위해 제안되었다, Scale-invariant한 특징을 제공하도록 설계된 pyramidal features 기반 backbone 구조는 object detection 및 classification 등에서 널리 쓰이고 있으며, 이러한 특징을 갖는 feature pyramid를 trainable attention mechanism에 적용하고자 할 때 계산량 및 메모리의 복잡도가 증가하는 문제가 있다. 본 논문에서는 일반적인 FPN에서 객체 인식 성능과 설명성을 높이기 위한 피라미드-주의집중 계층네트워크 (FPN-Attention Layered Network) 방식을 제안하고, 실험적으로 그 특성을 평가하고자 한다. 기존의 FPN만을 사용하였을 때 객체 인식 과정에서 설명성을 향상시키는 방식이 객체 인식에 미치는 정도를 정량적으로 평가하였다. 제안된 모델의 적용을 통해 낮은 computing 오버헤드 수준에서 multi-level feature를 고려한 시각적 설명성을 개선시켜, 결괴적으로 객체 인식 성능을 향상 시킬 수 있음을 실험적으로 확인할 수 있었다.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.16 no.5
• /
• pp.225-232
• /
• 2021

Deraining is one of the image restoration tasks and should consider a tradeoff between local details and broad contextual information while recovering images. Current studies adopt an attention mechanism which has been actively researched in natural language processing to deal with both global and local features. This paper classifies existing deraining methods and provides comparative analysis and performance comparison by using several datasets in terms of generalization.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.4
• /
• pp.980-997
• /
• 2024

To improve the brightness of images and reveal hidden information in dark areas is the main objective of low-light image enhancement (LLIE). LLIE methods based on deep learning show good performance. However, there are some limitations to these methods, such as the complex network model requires highly configurable environments, and deficient enhancement of edge details leads to blurring of the target content. Single-scale feature extraction results in the insufficient recovery of the hidden content of the enhanced images. This paper proposed an edge detection-based multi-scale feature enhancement network for LLIE (EDMFEN). To reduce the loss of edge details in the enhanced images, an edge extraction module consisting of a Sobel operator is introduced to obtain edge information by computing gradients of images. In addition, a multi-scale feature enhancement module (MSFEM) consisting of multi-scale feature extraction block (MSFEB) and a spatial attention mechanism is proposed to thoroughly recover the hidden content of the enhanced images and obtain richer features. Since the fused features may contain some useless information, the MSFEB is introduced so as to obtain the image features with different perceptual fields. To use the multi-scale features more effectively, a spatial attention mechanism module is used to retain the key features and improve the model performance after fusing multi-scale features. Experimental results on two datasets and five baseline datasets show that EDMFEN has good performance when compared with the stateof-the-art LLIE methods.

• Nuclear Engineering and Technology
• /
• v.56 no.6
• /
• pp.2343-2351
• /
• 2024

Accurately predicting the thermal hydraulic parameters of a transient reactor core under different working conditions is the first step toward reactor safety. Mass flow rate and temperature are important parameters of core thermal hydraulics, which have often been modeled as time series prediction problems. This study aims to achieve accurate and continuous prediction of core thermal hydraulic parameters under instantaneous conditions, as well as test the feasibility of a newly constructed gated recurrent unit (GRU) model based on the soft attention mechanism for core parameter predictions. Herein, the China Experimental Fast Reactor (CEFR) is used as the research object, and CEFR 1/2 core was taken as subject to carry out continuous predictive analysis of thermal parameters under transient conditions., while the subchannel analysis code named SUBCHANFLOW is used to generate the time series of core thermal-hydraulic parameters. The GRU model is used to predict the mass flow and temperature time series of the core. The results show that compared to the adaptive radial basis function neural network, the GRU network model produces better prediction results. The average relative error for temperature is less than 0.5 % when the step size is 3, and the prediction effect is better within 15 s. The average relative error of mass flow rate is less than 5 % when the step size is 10, and the prediction effect is better in the subsequent 12 s. The GRU model not only shows a higher prediction accuracy, but also captures the trends of the dynamic time series, which is useful for maintaining reactor safety and preventing nuclear power plant accidents. Furthermore, it can provide long-term continuous predictions under transient reactor conditions, which is useful for engineering applications and improving reactor safety.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.17 no.4
• /
• pp.190-197
• /
• 2024

The O2O store model dissolves the boundaries between online and offline platforms, providing significant convenience to customers. To effectively operate such platforms, small business owners must provide necessary information in digital format. Specifically, the process of digitizing Korean menus manually can lead to multiple issues, and the use of OCR technology often results in high error rates due to the low accuracy in recognizing Korean. In response, this paper proposes an enhanced OCR model based on the popular EasyOCR framework, aimed at improving the recognition accuracy of Korean. The proposed model integrates the structural advantages of VGG and ResNet, and incorporates an attention mechanism to significantly improve the recognition performance of Korean. Moreover, experimental results indicate that the proposed model achieved approximately a 3.5% improvement in accuracy and around a 1% improvement in both confidence score and normalized edit distance compared to EasyOCR. Therefore, this demonstrates that the proposed method effectively addresses the existing challenges.

• Journal of Korea Water Resources Association
• /
• v.55 no.7
• /
• pp.495-504
• /
• 2022

With the recent development of artificial intelligence, a Long Short-Term Memory (LSTM) model that is efficient with time-series analysis is being used to increase the accuracy of predicting the inflow of dams. In this study, we predict the inflow of the Soyang River dam, using the LSTM model with the Sequence-to-Sequence (LSTM-s2s) and attention mechanism (LSTM-s2s with attention) that can further improve the LSTM performance. Hourly inflow, temperature, and precipitation data from 2013 to 2020 were used to train the model, and validate and test for evaluating the performance of the models. As a result, the LSTM-s2s with attention showed better performance than the LSTM-s2s in general as well as in predicting a peak value. Both models captured the inflow pattern during the peaks but detailed hourly variability is limitedly simulated. We conclude that the proposed LSTM-s2s with attention can improve inflow forecasting despite its limits in hourly prediction.

• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.18 no.5
• /
• pp.521-528
• /
• 2009

In these days, biomimetic apprioach in the design and control of robotic system has drawn much attention. The human-being and mammals possess their own feet. Using the mobility of their feet, they are able to walk in various environments such as plain land, desert, swamp, and so on. Previously developed biped robots and four-legged robots did not employ such adaptable foot. In this work, a biomimetic foot mechanism is investigated through analysis of the foot structure of the human-being. This foot mechanism consists of a toe, an ankle, a heel, and some springs replacing the foot muscles and tendons. Using five toes and springs, this foot can adapt to various environments. A mathematical modeling for this foot mechanism was performed and its characteristics were observed through numerical simulation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.10
• /
• pp.4681-4702
• /
• 2016

Recently, Massive energy consumption in Cloud Storage System has attracted great attention both in industry and research community. However, most of the solutions utilize single method to reduce the energy consumption only in one aspect. This paper proposed an energy effective gear-shifting mechanism (E2GSM) in Cloud Storage System to save energy consumption from multi-aspects. E2GSM is established on data classification mechanism and data replication management strategy. Data is classified according to its properties and then be placed into the corresponding zones through the data classification mechanism. Data replication management strategies determine the minimum replica number through a mathematical model and make decision on replica placement. Based on the above data classification mechanism and replica management strategies, the energy effective gear-shifting mechanism (E2GSM) can automatically gear-shifting among the nodes. Mathematical analytical model certificates our proposed E2GSM is energy effective. Simulation experiments based on Gridsim show that the proposed gear-shifting mechanism is cost effective. Compared to the other energy-saved mechanism, our E2GSM can save energy consumption substantially at the slight expense of performance loss while meeting the QoS of user.

• Journal of the Korea Convergence Society
• /
• v.12 no.10
• /
• pp.55-61
• /
• 2021

In this paper, we proposed a DeepLabv3+ based encoder-decoder model utilizing an attention mechanism for precise semantic segmentation. The DeepLabv3+ is a semantic segmentation method based on deep learning and is mainly used in applications such as autonomous vehicles, and infrared image analysis. In the conventional DeepLabv3+, there is little use of the encoder's intermediate feature map in the decoder part, resulting in loss in restoration process. Such restoration loss causes a problem of reducing segmentation accuracy. Therefore, the proposed method firstly minimized the restoration loss by additionally using one intermediate feature map. Furthermore, we fused hierarchically from small feature map in order to effectively utilize this. Finally, we applied an attention mechanism to the decoder to maximize the decoder's ability to converge intermediate feature maps. We evaluated the proposed method on the Cityscapes dataset, which is commonly used for street scene image segmentation research. Experiment results showed that our proposed method improved segmentation results compared to the conventional DeepLabv3+. The proposed method can be used in applications that require high accuracy.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.34 no.4
• /
• pp.66-71
• /
• 2011

This article proposes a novel shift automation mechanism for an automated manual transmission (AMT). The development of an automated manual transmission is currently being paid considerable attention by vehicle manufacturers, with the prospects of combining the comfort of an automatic transmission and the high efficiency of a manual transmission. In order to automate the shift mechanism of a manual transmission, the proposed shift automation mechanism consists of two electric motors, cross shaped pinion gears, rack type shift rails, and a ball splined hollow shaft. First we describe the shift mechanism and operating principles of a manual transmission to investigate important design criteria for the shift automation device. And a new shift automation mechanism is described with its structure, elements, and operating principles in detail. Using a conventional manual transmission, we develop a full three-dimensional CAD model of an AMT which includes main components of the manual transmission and the designed shift automation mechanism. Finally we investigate the operating performances and feasibility of the designed AMT by a dynamic analysis.