• Annual Conference of KIPS
• /
• 2017.04a
• /
• pp.892-895
• /
• 2017

본 논문에서는 비분할 비디오로부터 이 비디오에 담긴 사람의 행동을 효과적으로 탐지해내기 위한 심층 신경망 모델을 제안한다. 일반적으로 비디오에서 사람의 행동을 탐지해내는 작업은 크게 비디오에서 행동 탐지에 효과적인 특징들을 추출해내는 과정과 이 특징들을 토대로 비디오에 담긴 행동을 탐지해내는 과정을 포함한다. 본 논문에서는 특징 추출 과정과 행동 탐지 과정에 이용할 심층 신경망 모델을 제시한다. 특히 비디오로부터 각 행동별 시간적, 공간적 패턴을 잘 표현할 수 있는 특징들을 추출해내기 위해서는 C3D 및 I-ResNet 합성곱 신경망 모델을 이용하고, 시계열 특징 벡터들로부터 행동을 자동 판별해내기 위해서는 양방향 BI-LSTM 순환 신경망 모델을 이용한다. 대용량의 공개 벤치 마크 데이터 집합인 ActivityNet 비디오 데이터를 이용한 실험을 통해, 본 논문에서 제안하는 심층 신경망 모델의 성능과 효과를 확인할 수 있었다.

• Journal of the Korea Convergence Society
• /
• v.9 no.4
• /
• pp.47-55
• /
• 2018

In this paper, we propose a novel face recognition(FR) method that takes advantage of combining weighted deep local features extracted from multiple Deep Convolutional Neural Networks(DCNNs) learned with a set of facial local regions. In the proposed method, the so-called weighed deep local features are generated from multiple DCNNs each trained with a particular face local region and the corresponding weight represents the importance of local region in terms of improving FR performance. Our weighted deep local features are applied to Joint Bayesian metric learning in conjunction with Nearest Neighbor(NN) Classifier for the purpose of FR. Systematic and comparative experiments show that our proposed method is robust to variations in pose, illumination, and expression. Also, experimental results demonstrate that our method is feasible for improving face recognition performance.

• The Korean Journal of Applied Statistics
• /
• v.31 no.2
• /
• pp.265-285
• /
• 2018

In this study, we applied a deep neural network model to predict four grades of fine dust $PM_{10}$, 'Good, Moderate, Bad, Very Bad' and two grades, 'Good or Moderate and Bad or Very Bad'. The deep neural network model and existing classification techniques (such as neural network model, multinomial logistic regression model, support vector machine, and random forest) were applied to fine dust daily data observed from 2010 to 2015 in six major metropolitan areas of Korea. Data analysis shows that the deep neural network model outperforms others in the sense of accuracy.

• Journal of Digital Convergence
• /
• v.16 no.12
• /
• pp.293-299
• /
• 2018

The weather has a lot of influence on the cultivation of crops. Weather information on agricultural crop cultivation areas is indispensable for efficient cultivation and management of agricultural crops. Despite the high demand for agricultural weather, research on this is in short supply. In this research, we deal with the production method of agricultural weather in Jeollanam-do, which is the main production area of onions through GloSea5 and deep learning. A deep neural network model using the sliding window method was used and utilized to train daily weather prediction for predicting the agricultural weather. RMSE and MAE are used for evaluating the accuracy of the model. The accuracy improves as the learning period increases, so we compare the prediction performance according to the learning period and the prediction period. As a result of the analysis, although the learning period and the prediction period are similar, there was a limit to reflect the trend according to the seasonal change. a modified deep layer neural network model was presented, that applying the difference between the predicted value and the observed value to the next day predicted value.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.3
• /
• pp.308-313
• /
• 2019

Parkinson's disease, one of the three major diseases in old age, has more than 70 % of patients with speech disorders, and recently, diagnostic methods of Parkinson's disease through speech signals have been devised. In this paper, we propose a method of diagnosis of Parkinson's disease based on deep residual gated recurrent neural network using speech features. In the proposed method, the speech features for diagnosing Parkinson's disease are selected and applied to the deep residual gated recurrent neural network to classify Parkinson's disease patients. The proposed deep residual gated recurrent neural network, an algorithm combining residual learning with deep gated recurrent neural network, has a higher recognition rate than the traditional method in Parkinson's disease diagnosis.

• The Journal of Korean Institute of Next Generation Computing
• /
• v.14 no.6
• /
• pp.87-97
• /
• 2018

Rainfall depth is an important meteorological information. Generally, high spatial resolution rainfall data such as road-level rainfall data are more beneficial. However, it is expensive to set up sufficient Automatic Weather Systems to get the road-level rainfall data. In this paper, we proposed to use deep learning to recognize rainfall depth from road surveillance videos. To achieve this goal, we collected two new video datasets, and proposed a new deep learning architecture named Temporal and Spatial Segment Networks (TSSN) for rainfall depth recognition. Under TSSN, the experimental results show that the combination of the video frame and the differential frame is a superior solution for the rainfall depth recognition. Also, the proposed TSSN architecture outperforms other architectures implemented in this paper.

• Journal of Korea Water Resources Association
• /
• v.53 no.12
• /
• pp.1159-1172
• /
• 2020

This study applied deep convolution neural network based on U-Net and SegNet using long period weather radar data to very short-term rainfall prediction. And the results were compared and evaluated with the translation model. For training and validation of deep neural network, Mt. Gwanak and Mt. Gwangdeoksan radar data were collected from 2010 to 2016 and converted to a gray-scale image file in an HDF5 format with a 1km spatial resolution. The deep neural network model was trained to predict precipitation after 10 minutes by using the four consecutive radar image data, and the recursive method of repeating forecasts was applied to carry out lead time 60 minutes with the pretrained deep neural network model. To evaluate the performance of deep neural network prediction model, 24 rain cases in 2017 were forecast for rainfall up to 60 minutes in advance. As a result of evaluating the predicted performance by calculating the mean absolute error (MAE) and critical success index (CSI) at the threshold of 0.1, 1, and 5 mm/hr, the deep neural network model showed better performance in the case of rainfall threshold of 0.1, 1 mm/hr in terms of MAE, and showed better performance than the translation model for lead time 50 minutes in terms of CSI. In particular, although the deep neural network prediction model performed generally better than the translation model for weak rainfall of 5 mm/hr or less, the deep neural network prediction model had limitations in predicting distinct precipitation characteristics of high intensity as a result of the evaluation of threshold of 5 mm/hr. The longer lead time, the spatial smoothness increase with lead time thereby reducing the accuracy of rainfall prediction The translation model turned out to be superior in predicting the exceedance of higher intensity thresholds (> 5 mm/hr) because it preserves distinct precipitation characteristics, but the rainfall position tends to shift incorrectly. This study are expected to be helpful for the improvement of radar rainfall prediction model using deep neural networks in the future. In addition, the massive weather radar data established in this study will be provided through open repositories for future use in subsequent studies.

• Journal of the Korea Computer Graphics Society
• /
• v.30 no.4
• /
• pp.19-29
• /
• 2024

In this paper, we examine the effectiveness of StyleGAN-generated images for data augmentation in training deep neural networks for medical image classification. We apply StyleGAN data augmentation to train VGG-16 networks for pneumonia diagnosis from chest X-ray images and focal liver lesion classification from abdominal CT images. Through quantitative and qualitative analyses, our experiments reveal that StyleGAN data augmentation expands the outer class boundaries in the feature space. Thanks to this expansion characteristics, the StyleGAN data augmentation can enhance classification performance when properly combined with real training images.

• Broadcasting and Media Magazine
• /
• v.25 no.2
• /
• pp.52-60
• /
• 2020

최근 생체 정보를 이용한 사용자 인증 기술이 발전하면서 이를 모바일 기기에 적용하는 사례가 크게 증가하고 있다. 특히, 얼굴 기반 인증 방식은 비접촉식이며 사용이 편리하여 적용 범위가 점점 확대되고 있는 추세이다. 그러나, 사용자의 얼굴 사진이나 동영상 등을 이용한 위변조가 용이하기 때문에 모바일 기기 내 보안 유지에 어려움을 야기한다. 본 고에서는 이러한 문제를 해결하기 위해 최근 활발히 연구되고 있는 심층신경망 기반 얼굴 위변조 검출 연구의 최신 동향을 소개하고자 한다. 먼저, 기본 합성곱 신경망 구조부터 생성모델 기반의 위변조 검출 방법까지 다양한 신경망 구조를 이용한 위변조 검출 방법에 대해 설명한다. 또한, 심층신경망 학습을 위해 사용되는 얼굴 위변조 데이터셋에 대해서도 간략히 살펴보고자 한다.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.48 no.10
• /
• pp.773-782
• /
• 2020

The development of joint FE models for deep learning neural network (DLNN)-based hybrid FEA is presented. Material models of bolts and bearings in the front axle of tractor, showing complex behavior induced by various tightening conditions, were replaced with DLNN models. Bolts are modeled as one-dimensional Timoshenko beam elements with six degrees of freedom, and bearings as three-dimensional solid elements. Stress-strain data were extracted from all elements after finite element analysis subjected to various load conditions, and DLNN for bolts and bearing were trained with Tensorflow. The DLNN-based joint models were implemented in the ABAQUS user subroutines where stresses from the next increment are updated and the algorithmic tangent stiffness matrix is calculated. Generalization of the trained DLNN in the FE model was verified by subjecting it to a new loading condition. Finally, the DLNN-based FEA for the front axle of the tractor was conducted and the feasibility was verified by comparing with results of a static structural experiment of the actual tractor.