• Journal of Convergence for Information Technology
• /
• v.11 no.9
• /
• pp.28-34
• /
• 2021

In the case of object detection using deep learning, both accuracy and real-time are required. However, it is difficult to use a deep learning model that processes a large amount of data in a limited resource environment. To solve this problem, this paper proposes an object detection model for small embedded devices. Unlike the general detection model, the model size was minimized by using a structure in which the pre-trained feature extractor was removed. The structure of the model was designed by repeatedly stacking lightweight convolution blocks. In addition, the number of region proposals is greatly reduced to reduce detection overhead. The proposed model was trained and evaluated using the public dataset PASCAL VOC. For quantitative evaluation of the model, detection performance was measured with average precision used in the detection field. And the detection speed was measured in a Raspberry Pi similar to an actual embedded device. Through the experiment, we achieved improved accuracy and faster reasoning speed compared to the existing detection method.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.14 no.5
• /
• pp.969-976
• /
• 2019

A lightweight artificial intelligence hardware has made great strides in many application areas. In general, a lightweight artificial intelligence system consist of lightweight artificial intelligence engine and preprocessor including feature selection, generation, extraction, and normalization. In order to achieve optimal performance in broad range of applications, lightweight artificial intelligence system needs to choose a good preprocessing function and set their respective hyper-parameters. This paper proposes a unified framework for a lightweight artificial intelligence system and utilization method for finding models with optimal performance to use on a given dataset. The proposed unified framework can easily generate a model combined with preprocessing functions and lightweight artificial intelligence engine. In performance evaluation using handwritten image dataset and fall detection dataset measured with inertial sensor, the proposed unified framework showed building optimal artificial intelligence models with over 90% test accuracy.

• Journal of the Korean Society for Precision Engineering
• /
• v.20 no.7
• /
• pp.177-184
• /
• 2003

In this study we performed optimal design for the vehicle mechanical component which satisfies both a sufficient stiffness and a lightweight using topology optimization technique. The FEA for the initial model before optimal design is performed by ABAQUS/Standard. And, we suggest optimization model using the topology optimal design program Altair Optisturuct 3.6. The FEA of optimal design is performed under the same condition as the initial model. We performed the FEA fur the topology optimal design model and verified the validity of the present method.

• Structural Engineering and Mechanics
• /
• v.50 no.5
• /
• pp.635-652
• /
• 2014

Creep and shrinkage behaviour of an ultra lightweight cement composite (ULCC) up to 450 days was evaluated in comparison with those of a normal weight aggregate concrete (NWAC) and a lightweight aggregate concrete (LWAC) with similar 28-day compressive strength. The ULCC is characterized by low density < 1500 $kg/m^3$ and high compressive strength about 60 MPa. Autogenous shrinkage increased rapidly in the ULCC at early-age and almost 95% occurred prior to the start of creep test at 28 days. Hence, majority of shrinkage of the ULCC during creep test was drying shrinkage. Total shrinkage of the ULCC during the 450-day creep test was the lowest compared to the NWAC and LWAC. However, corresponding total creep in the ULCC was the highest with high proportion attributed to basic creep (${\geq}$ ~90%) and limited drying creep. The high creep of the ULCC is likely due to its low elastic modulus. Specific creep of the ULCC was similar to that of the NWAC, but more than 80% higher than the LWAC. Creep coefficient of the ULCC was about 47% lower than that of the NWAC but about 18% higher than that of the LWAC. Among five creep models evaluated which tend to over-estimate the creep coefficient of the ULCC, EC2 model gives acceptable prediction within +25% deviations. The EC2 model may be used as a first approximate for the creep of ULCC in the designs of steel-concrete composites or sandwich structures in the absence of other relevant creep data.

• Journal of the Korea Concrete Institute
• /
• v.20 no.3
• /
• pp.405-412
• /
• 2008

Six alkali-activated (AA) concrete mixes were tested to explore the significance and limitations of developing an environmental friendly concrete. Ground granulated blast-furnace slag and powder typed sodium silicate were selected as source material and an alkaline activator, respectively. The main parameter investigated was the replacement level of lightweight fine aggregate to the natural sand. Workability and mechanical properties of lightweight AA concrete were measured: the variation of slump with time, the rate of compressive strength development, the splitting tensile strength, the moduli of rupture and elasticity, the stress-strain relationship, the bond resistance and shrinkage strain. Test results showed that the compressive strength of lightweight AA concrete sharply decreased when the replacement level of lightweight fine aggregate exceeded 30%. In particular, the increase in the discontinuous grading of lightweight aggregate resulted in the deterioration of the mechanical properties of concrete tested. The measured properties of lightweight AA concrete were also compared, wherever possible, with the results obtained from the design equations specified in ACI 318-05 or EC 2, depending on the relevance, and the results predicted from the empirical equations proposed by Slate et al. for lightweight ordinary Portland cement concrete. The stress-strain curves of different concrete were compared with predictions obtained from the mathematical model proposed by Tasnimi. The measured mechanical properties of lightweight AA concrete generally showed little agreement with the predictions obtained from these equations.

• Journal of Aerospace System Engineering
• /
• v.7 no.2
• /
• pp.8-14
• /
• 2013

When launch vehicles are manufactured, one of the key points is a design of lightweight structure for reducing costs. Isogrid structure was designed to solve this topic, and many researches were carried out about buckling load because compression load is mainly applied to them. Recently, many studies are also being carried out about FEM model geometry of isogrid structure. The reason is that isogrid structure depends on size of ribs so it is difficult to modify about small changes in rib pattern. In this study, 1/8 model of cylindrical isogrid structure model was developed to analyze buckling behavior. Through parameter study, buckling analysis were performed to analyze buckling load and buckling mode depending on size of ribs.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2012.05a
• /
• pp.827-828
• /
• 2012

WG-7 is a stream cipher based on WG Stream Cipher and is designed by Y. Luo, Q. Chai, G. Gong, and X. Lai in 2010. This cipher is designed to implement in low cost and lightweight application such as RFID tags. In this paper, we survey and compare cryptographic module such as stream and block cipher. We can estimate security performance suitable to system.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.2
• /
• pp.51-58
• /
• 2023

This paper proposes a deep learning method for estimating the heart rate from facial videos. Our proposed method estimates remote photoplethysmography (rPPG) signals to predict the heart rate. Although there have been proposed several methods for estimating rPPG signals, most previous methods can not be utilized in low-power single board computers due to their computational complexity. To address this problem, we construct a lightweight student model and employ a knowledge distillation technique to reduce the performance degradation of a deeper network model. The teacher model consists of 795k parameters, whereas the student model only contains 24k parameters, and therefore, the inference time was reduced with the factor of 10. By distilling the knowledge of the intermediate feature maps of the teacher model, we improved the accuracy of the student model for estimating the heart rate. Experiments were conducted on the UBFC-rPPG dataset to demonstrate the effectiveness of the proposed method. Moreover, we collected our own dataset to verify the accuracy and processing time of the proposed method on a real-world dataset. Experimental results on a NVIDIA Jetson Nano board demonstrate that our proposed method can infer the heart rate in real time with the mean absolute error of 2.5183 bpm.

• Computational Structural Engineering
• /
• v.36 no.2
• /
• pp.19-23
• /
• 2023

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.10 no.3
• /
• pp.589-595
• /
• 2009

This paper proposes an object-oriented software development guidance and an evaluation index for the productivity related to Spring Framework 2.5. Spring Framework is a known successful open source standard model for lightweight container architecture. Non EJB and the EBJ architecture to resolve the problem with benefits to support the new architecture is a lightweight container architecture. This architecture, such as the EJB, but not heavy, to provide all of the architecture is possible. The lightweight container architecture is most often used in business spring framework is well-known architecture. Therefore, this research has the Non EJB and the EJB to solve the advantages and disadvantages developed to support the latest spring framework 2.5 lightweight container architecture based on the design and implementation of a hotel reservation system with the objective through the specification of the software previously to provide guidance to development productivity.