• Transactions on Semiconductor Engineering
• /
• v.2 no.1
• /
• pp.17-20
• /
• 2024

This paper presents a novel lightweight object detection model tailored for low-powered edge devices, addressing the limitations of traditional resource-intensive computer vision models. Our proposed detector, inspired by the Single Shot Detector (SSD), employs a compact yet robust network design. Crucially, it integrates an 'enhancer block' that significantly boosts its efficiency in detecting smaller objects. The model comprises two primary components: the Light_Block for efficient feature extraction using Depth-wise and Pointwise Convolution layers, and the Enhancer_Block for enhanced detection of tiny objects. Trained from scratch on the Udacity Annotated Dataset with image dimensions of 300x480, our model eschews the need for pre-trained classification weights. Weighing only 5.5MB with approximately 0.43M parameters, our detector achieved a mean average precision (mAP) of 27.7% and processed at 140 FPS, outperforming conventional models in both precision and efficiency. This research underscores the potential of lightweight designs in advancing object detection for edge devices without compromising accuracy.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.11 no.1
• /
• pp.39-46
• /
• 2016

According to general architecture of flight simulator made up of several independent rendering display systems, display synchronization problem between them naturally happens. In addition, since the flight simulator is usually implemented in the same networks where network delay is not big concern, it is necessary to consider different factors of existing synchronization technique. Among them, in this paper, we propose a new display synchronization scheme for flight simulator where each system has different rendering capacity. In the proposed scheme, each system is synchronized by considering maximum and minimum frames per second (FPS) while considering level of detail and latency. Also, experimental results are given to demonstrate the suitability of the proposed scheme for display synchronization scheme.

• Journal of the Korean Society of Visualization
• /
• v.9 no.4
• /
• pp.63-68
• /
• 2011

The flow-field of a liquid-metal system is very important for the safety analysis and the design of the steam generator of liquid-metal fast breeder reactor. Dynamic neutron radiography (DNR) is suitable for a visualization and measurement of a liquid metal flow and a two-phase flow in a metallic duct. However, the three dimensional DNR techniques is not enough to obtain the velocity information in the wide channel up to now. In this research, a high speed DNR technique was applied to visualize the heavy liquid-metal flow field in the narrow channel with the HANARO-beam facility. The images were taken with a high frame-rate neutron radiography at 250 fps and analyzed with a Particle Image Velocimetry(PIV) method. The images were compared with the results of the commercial CFX code to study the feasibility of DNR technique for the measuring the heavy liquid-metal flow field. The PIV images could discern the turbulent vortex flow in the two-dimensional narrow channel.

• ETRI Journal
• /
• v.30 no.1
• /
• pp.167-169
• /
• 2008

In this paper, we propose hardware architecture for a high-speed context-adaptive variable length coding (CAVLC) decoder in H.264. In the CAVLC decoder, the codeword length of the current decoding block is used to determine the next input bitstreams (valid bits). Since the computation of valid bits increases the total processing time of CAVLC, we propose two techniques to reduce processing time: one is to reduce the number of decoding steps by introducing a lookup table, and the other is to reduce cycles for calculating the valid bits. The proposed CAVLC decoder can decode $1920{\times}1088$ 30 fps video in real time at a 30.8 MHz clock.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.12 no.2
• /
• pp.127-138
• /
• 2012

In this paper, we propose a real-time virtual re-convergence hardware platform especially to reduce the visual fatigue caused by stereoscopy. Our unique idea to reduce visual fatigue is to utilize the virtual re-convergence based on the optimized disparity-map that contains more depth information in the negative disparity area than in the positive area. Our virtual re-convergence hardware platform, which consists of image rectification, disparity estimation, depth post-processing, and virtual view control, is realized in real time with 60 fps on a single Xilinx Virtex-5 FPGA chip.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.12 no.2
• /
• pp.203-211
• /
• 2012

This paper proposes a $16{\times}16$ and $32{\times}32$ inverse transform architecture for HEVC (High Efficiency Video Coding). HEVC large transform of $16{\times}16$ and $32{\times}32$ suffers from huge computational complexity. To resolve this problem, we proposed a new large inverse transform architecture based on hardware reuse. The processing element is optimized by exploiting fully recursive and regular butterfly structure. To achieve low area, the processing element is implemented by shifters and adders without multiplier. Implementation of the proposed 2-D inverse transform architecture in 0.18 ${\mu}m$ technology shows about 300 MHz frequency and 287 Kgates area, which can process 4K ($3840{\times}2160$)@ 30 fps image.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2016.07a
• /
• pp.27-28
• /
• 2016

실시간 렌더링 시스템에서 시야를 변경하는 속도가 증가할수록 FPS(Frame Per Second)가 감소하는 문제가 발생한다. 이에 따라 시야 변경이 증가할수록 감소하는 FPS를 향상하기 위한 렌더링 최적화 기술이 요구되며, 본 논문에서는 동적 LOD(Level of Detail)와 컬링 기술을 적용하여 실시간 렌더링 과정에서 렌더링 데이터를 감소함으로써 저하되는 FPS를 증가시키기 위한 렌더링 최적화 기법과 HMD(Helmet Mounted Display)의 움직임에 따라 변경된 시야각을 적용한 OSG(Open Scene Graph) 기반 실시간 렌더링 시스템을 제시한다.

• Proceedings of the KSME Conference
• /
• 2001.06e
• /
• pp.63-68
• /
• 2001

An experimental study was carried out to identify the various regimes of natural convective boiling and to determine the Critical Heat Flux(CHF) on a 70mm square surface which is inclined at $180^{\circ}$(upward), $90^{\circ}, \;45^{\circ}$. The heater block made of copper with cartridge heaters is submerged in a water tank with windows for visualization. As the heat flux increases from $100kW/m^2$ to $1.1MW/m^2$, the heat transfer regime migrates from the nucleate boiling to film boiling and results in a rapid heat up of the heater block. An explosive vapor generation on the heated surface, whose size and frequency are characterized by the heat flux, is visualized by using a digital camcorder with $512{\times}512$ pixel size at 30fps.

• Proceedings of the Korean Society For Composite Materials Conference
• /
• 2002.05a
• /
• pp.173-176
• /
• 2002

This paper describes the design procedure of composite parts for hingeless hub system of helicopter. These composite parts are composed of flexure and sleeve. The design of this composite flexure is an important technique in hingeless hub system since the rotor system stability and dynamic characteristics depends on this flexure characteristics. The objective of this research is to replace the existing metal hub parts with composite to improve the performance and stability. First, the coupon test of candidate composite material for hub parts was conducted. The hub parts was designed based on test results and the manufacturing possibility by using Fiber Placement System(FPS) was checked. Also the dynamic analysis and stress analysis of composite hub parts was conducted. Through this research, we will find out the possibility of replacing existing metal hub parts with composite.

• Journal of Ocean Engineering and Technology
• /
• v.32 no.2
• /
• pp.143-150
• /
• 2018

In this study, we designed and developed an underwater LED communication system composed of an LED and a photo sensor. In addition, we experimented with video data transmission in a water tank. Two communication modules were installed in the 3 m water tank, and the image data transmission test was successfully performed at a rate of 20 frames per second(FPS), image resolution of $480{\times}272$, and data communication speed of 4 Mbps.