• Journal of Korean Ophthalmic Optics Society
• /
• v.6 no.1
• /
• pp.87-91
• /
• 2001

The a-C:H films have been grown on the glass substrate by PECVD method, where plasma was generated with a 60 Hz line power source. The carbonization is checked from peak intensities of D($sp^3$) and G($sp^2$) peaks in Raman spectra and is analyzed using the Gaussian convolution method of spectrum. Both the bonding strength of C-H and the ratio of $sp^3$ to $sp^2$ in bonding are found to be slightly dependent of partial pressure of $C_2H_2$.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.06a
• /
• pp.84-85
• /
• 2019

MPEG-NNR (Compression of Neural Network for Multimedia Content Description and Analysis) aims to define a compressed and interoperable representation of trained neural networks. In this paper, we present a low-rank approximation to compress a CNN used for image compression, which is one of MPEG-NNR use cases. In the presented method, the low-rank approximation decomposes one 2D kernel matrix of weights into two 1D kernel matrix values in each convolution layer to reduce the data amount of weights. The evaluation results show that the model size of the original CNN is reduced to half as well as the inference runtime is reduced up to about 30% with negligible loss in PSNR.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.01a
• /
• pp.369-372
• /
• 2023

본 논문에서는 삼각형 구조로 구성된 3차원 메쉬(Mesh)에서 합성곱 신경망(Convolution Neural Network, CNN)을 응용하여 정확도가 높은 새로운 학습 표현 기법을 제시한다. 우리는 메쉬를 구성하고 있는 폴리곤의 edge와 face의 로컬 특징을 기반으로 학습을 진행한다. 일반적으로 딥러닝은 인공신경망을 수많은 계층 형태로 연결한 기법을 말하며, 주요 처리 대상은 1, 2차원 데이터 형태인 오디오 파일과 이미지였다. 인공지능에 대한 연구가 지속되면서 3차원 딥러닝이 도입되었지만, 기존의 학습과는 달리 3차원 딥러닝은 데이터의 확보가 쉽지 않다. 혼합현실과 메타버스 시장의 확대로 인해 3차원 모델링 시장이 증가하고, 기술의 발전으로 데이터를 획득할 수 있는 방법이 생겼지만, 3차원 데이터를 직접적으로 학습에 이용하는 방식으로 적용하는 것은 쉽지 않다. 그렇게 때문에 본 논문에서는 산업 현장에서 이용되는 데이터인 메쉬 구조를 폴리곤의 최소 단위인 삼각형 형태로 구성하여 학습 데이터를 구성해 기존의 방법보다 정확도가 높은 학습 기법을 제안한다.

• The Bulletin of The Korean Astronomical Society
• /
• v.42 no.2
• /
• pp.68-68
• /
• 2017

Measurements of solar wind speed near the Sun (< 0.1 AU) are important for understanding acceleration mechanism of solar wind as well as space weather predictions, but hard to directly measure them. For the first time, we provide 2D solar wind speed maps in the LASCO field of view using three consecutive days data. By applying the Fourier convolution and inverse Fourier transform, we decompose the 3D intensity data (r, PA, t) into the 4D one (r, PA, t, v). Then, we take the weighted mean along speed to determine the solar wind speeds that gives V(r, PA, t) in every 30 min. The estimated radial speeds are consistent with those given by an artificial flow and plasma blobs. We find that the estimated speeds are moderately correlated with those from slow CMEs and those from IPS observations. A comparison of yearly solar wind speed maps in 2000 and 2009 shows that they have very remarkable differences: azimuthally uniform distribution in 2000 and bi-modal distribution (high speed near the poles and low speed near the equator) in 2009.

• Journal of the Semiconductor & Display Technology
• /
• v.21 no.2
• /
• pp.68-73
• /
• 2022

In this study, we produced the surface defect detection model to automatically detect defect bone plates using a deep learning algorithm. Bone plates with a width and a length of 50 mm are most used for fracture treatment. Normal bone plates and defective bone plates were printed on the 3d printer. Normal bone plates and defective bone plates were photographed with 1,080 pixels using the webcam. The total quantity of collected images was 500. 300 images were used to learn the defect detection model. 200 images were used to test the defect detection model. The mAP(Mean Average Precision) method was used to evaluate the performance of the surface defect detection model. As the result of confirming the performance of the surface defect detection model, the detection accuracy was 96.3 %.

• Electronics and Telecommunications Trends
• /
• v.34 no.4
• /
• pp.15-22
• /
• 2019

Recent technological advances in three-dimensional (3D) sensing devices and machine learning such as deep leaning has enabled data-driven 3D applications. Research on artificial intelligence has developed for the past few years and 3D deep learning has been introduced. This is the result of the availability of high-quality big data, increases in computing power, and development of new algorithms; before the introduction of 3D deep leaning, the main targets for deep learning were one-dimensional (1D) audio files and two-dimensional (2D) images. The research field of deep leaning has extended from discriminative models such as classification/segmentation/reconstruction models to generative models such as those including style transfer and generation of non-existing data. Unlike 2D learning, it is not easy to acquire 3D learning data. Although low-cost 3D data acquisition sensors have become increasingly popular owing to advances in 3D vision technology, the generation/acquisition of 3D data is still very difficult. Even if 3D data can be acquired, post-processing remains a significant problem. Moreover, it is not easy to directly apply existing network models such as convolution networks owing to the various ways in which 3D data is represented. In this paper, we summarize technological trends in AI-based 3D content generation.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2022.06a
• /
• pp.110-110
• /
• 2022

본 연구는 자율운항 선박의 원격 고장 진단 기법 개발의 일부로 수행되었다. 특히, 엔진 연료 계통 장비로부터 계측된 시계열 데이터로부터 상태 진단을 위한 알고리즘 구현 결과를 제시하였다. 엔진 연료 펌프와 청정기를 가진 육상 실험 장비로부터 진동 시계열 데이터 계측하였으며, 이상 감지, 고장 분류 및 고장 예측이 가능한 심층 학습(Deep Learning) 및 기계 학습(Machine Learning) 알고리즘을 구현하였다. 육상 실험 장비에 고장 유형 별로 인위적인 고장을 발생시켜 특징적인 진동 신호를 계측하여, 인공 지능 학습에 이용하였다. 계측된 신호 데이터는 선행 발생한 사건의 신호가 후행 사건에 영향을 미치는 특성을 가지고 있으므로, 시계열에 내포된 고장 상태는 시간 간의 선후 종속성을 반영할 수 있는 학습 알고리즘을 제시하였다. 고장 사건의 시간 종속성을 반영할 수 있도록 순환(Recurrent) 계열의 RNN(Recurrent Neural Networks), LSTM(Long Short-Term Memory models)의 모델과 합성곱 연산 (Convolution Neural Network)을 기반으로 하는 Conv1D 모델을 적용하여 예측 정확성을 비교하였다. 특히, 합성곱 계열의 RNN LSTM 모델이 고차원의 순차적 자연어 언어 처리에 장점을 보이는 모델임을 착안하여, 신호의 시간 종속성을 학습에 반영할 수 있는 합성곱 계열의 Conv1 알고리즘을 고장 예측에 사용하였다. 또한 기계 학습 모델의 효율성을 감안하여 XGBoost를 추가로 적용하여 고장 예측을 시도하였다. 최종적으로 연료 펌프와 청정기의 진동 신호로부터 Conv1D 모델과 XGBoost 모델의 고장 예측 성능 결과를 비교하였다

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.7
• /
• pp.29-38
• /
• 2009

In this paper, a high performance pipelined computing design of parallel multiplier-temporal buffer-parallel accumulator is present for the convolution-based non-cascaded architecture aiming at the real time Discrete Wavelet Transform(DWT) processing. The convolved multiplication of DWT would be reduced upto 1/4 by utilizing the filter coefficients symmetry and the up/down sampling; and it could be dealt with 3-5 times faster computation by LUT-based DA multiplication of multiple filter coefficients parallelized for product terms with an image data. Further, the reutilization of computed product terms could be achieved by storing in the temporal buffer, which yields the saving of computation as well as dynamic power by 50%. The convolved product terms of image data and filter coefficients are realigned and stored in the temporal buffer for the accumulated addition. Then, the buffer management of parallel aligned storage is carried out for the high speed sequential retrieval of parallel accumulations. The convolved computation is pipelined with parallel multiplier-temporal buffer-parallel accumulation in which the parallelization of temporal buffer and accumulator is optimize, with respect to the performance of parallel DA multiplier, to improve the pipelining performance. The proposed architecture is back-end designed with 0.18um library, which verifies the 30fps throughput of SVGA(800$\times$600) images at 90MHz.

• Journal of the Korea Computer Graphics Society
• /
• v.26 no.2
• /
• pp.11-19
• /
• 2020

When a character's pose changes, its center of mass(COM) also changes. The change of COM has distinctive patterns corresponding to various motion types like walking, running or sitting. Thus the motion type can be predicted by using COM movement. We propose a motion generator that uses character's center of mass information. This generator can generate various motions without annotated action type labels. Thus dataset for training and running can be generated full-automatically. Our neural network model takes the motion history of the character and its center of mass information as inputs and generates a full-body pose for the current frame, and is trained using simple Convolutional Neural Network(CNN) that performs 1D convolution to deal with time-series motion data.

• Journal of Institute of Convergence Technology
• /
• v.1 no.1
• /
• pp.18-23
• /
• 2011

Shaping an input command through considering the resonant modes of multi degrees of freedom system, it is possible to realize the wanted motion, without exciting the uncontrollable modes of the flexible system. But, an increase of modes to be considered brings inevitably about the time delay due to an excessive rising time. On the purpose of reducing the rising time, only the interesting and dominant modes can be considered to determine the timing pulses of input shaper. In this paper, an effect of shaper by the partial modes is analysed for a specific system and the input shapers by the partial modes are analysed for three d.o.f damped system, using Matlab simulation.