• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.325-327
• /
• 2021

코로나19 전염병 예방을 위한 공공장소에서의 마스크 착용이 의무화되고 있다. 그러나 사람들이 다양한 이유로 마스크를 제대로 착용하지 않아 감염에 노출되는 위험이 발생하고 있다. 이러한 방역 문제를 해결하고 본 논문은 영상을 인식하여 마스크를 쓴 얼굴과 쓰지 않은 얼굴을 검출하는 방식을 제안한다. 제안 방법은 마스크 착용자와 비착용자 얼굴 영상을 딥러닝 기반의 YOLO 네트워크로 학습하여, 마스크 착용 유무를 판별한다. 동일 YOLO 네트워크에 대해 여러가지 조건으로 학습을 수행하고, 학습에 사용되지 않은 검증 데이터를 이용해 정확도가 가장 높은 네트워크의 가중치를 선택하였다. 실험결과, 마스크 착용자는 67.2%, 미착용자는 39.8%의 판별 정확도를 보였다. 미착용자에 대해 낮은 정확도를 보인 이유는 학습 데이터의 부족으로 판단되며, 이를 보완하기 위하여 더 많은 학습데이터를 제작하여 성능을 개선시키고자 한다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.686-688
• /
• 2020

본 논문에서는 U-Net 모델을 이용하여 정교하고 반복되는 패턴을 가진 인쇄물에 대한 비지도 학습을 통한 딥러닝 기반 이상치탐지(Anomaly Detection) 방법을 제안하였다. 인쇄물(카드)의 비정상 패턴 검출을 위하여 촬영한 영상으로부터 카드 영역을 분리한 이미지로 구성된 Dataset을 구축하였고 정상 이미지와 동일한 이미지를 출력하기 위해, 정상 이미지와 마스크 이미지 쌍의 Training dataset을 U-Net으로 학습하였다. Test dataset의 이미지를 입력으로 넣어 생성된 마스크 결과를 원본 마스크 이미지와 비교하여 이상 여부를 판단하는 본 논문의 방법이 정상, 비정상 인쇄물을 잘 구분하는 것을 확인하였다. 또한 정상과 비정상 이미지 각각을 학습한 지도학습 기반 CNN 분류 방법을 입력 영상과 복원 영상 간의 복원 오차를 비교하여 객체의 이상 여부를 판별하는 본 논문의 방법과 비교 평가하였다. 본 논문을 통해 U-Net을 사용하여 별도로 데이터에 대한 label 취득 없이 이상치를 검출할 수 있음을 확인할 수 있었다.

• KIISE Transactions on Computing Practices
• /
• v.24 no.1
• /
• pp.53-57
• /
• 2018

A poisoning accident is often caused by a situation in which people eat poisonous mushrooms because they cannot distinguish between edible mushrooms and poisonous mushrooms. In this paper, we propose an automatic mushroom recognition system by using the convolutional neural network. We collected 1478 mushroom images of 38 species using image crawling, and used the dataset for learning the convolutional neural network. A comparison experiment using AlexNet, VGGNet, and GoogLeNet was performed using the collected datasets, and a comparison experiment using a class number expansion and a fine-tuning technique for transfer learning were performed. As a result of our experiment, we achieve 82.63% top-1 accuracy and 96.84% top-5 accuracy on test set of our dataset.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.2
• /
• pp.569-574
• /
• 2023

For maintenance of concrete structures, it is necessary to identify and maintain various defects. With the current method, there are problems with efficiency, safety, and reliability when inspecting large-scale social infrastructure, so it is necessary to introduce a new inspection method. Recently, with the development of deep learning technology for images, concrete defect classification research is being actively conducted. However, studies on contamination and spalling other than cracks are limited. In this study, a variety of concrete defect type classification models were developed through transfer learning on a pre-learned deep learning model, factors that reduce accuracy were derived, and future development directions were presented. This is expected to be highly utilized in the field of concrete maintenance in the future.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.07a
• /
• pp.7-10
• /
• 2021

에스컬레이터의 고장 여부를 사전에 파악하는 것은 경제적 손실뿐만 아니라 인명 피해를 예방할 수 있어서 매우 중요하다. 실제 이러한 고장 예측을 위한 많은 딥러닝 알고리즘이 연구되고 있지만, 설비의 이상 데이터 확보가 어려워 모델 학습이 어렵다는 문제점이 있다. 본 연구에서는 이러한 문제의 해결 방안으로 비지도 학습 기반의 방법론 중 하나인 LSTM Autoencoder 알고리즘을 사용해 에스컬레이터의 이상을 탐지하는 모델을 생성했고, 최종 실험 결과 모델 성능 AUROC가 0.9966, 테스트 Accuracy가 0.97이라는 높은 정확도를 기록했다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.3
• /
• pp.355-363
• /
• 2020

In this paper, we propose a deep learning training method for accurately classifying Kuzushiji-MNIST and Kuzushiji-49 datasets for ancient and medieval Japanese characters. We analyze the latest convolutional neural network networks through experiments to select the most suitable network, and then use the networks to select the number of training to classify Kuzushiji-MNIST and Kuzushiji-49 datasets. In addition, the training is conducted with high accuracy by applying learning methods such as Mixup and Random Erase. As a result of the training, the accuracy of the proposed method can be shown to be high by 99.75% for MNIST, 99.07% for Kuzushiji-MNIST, and 97.56% for Kuzushiji-49. Through this deep learning-based technology, it is thought to provide a good research base for various researchers who study East Asian and Western history, literature, and culture.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.353-353
• /
• 2022

본 연구에서는 CMIP5(The fifth phase of the Couple Model Intercomparison Project) 미래기후시나리오와 LSTM(Long Short-Term Memory) 모형 기반의 딥러닝 기법을 이용하여 하천유량 예측을 위한 최적 학습 기간을 제시하였다. 연구지역으로는 진안군(성산리) 지점을 선정하였다. 보정(2000~2002/2014~2015) 및 검증(2003~2005/2016~2017) 기간을 설정하여 연구지역의 실측 유량 자료와 LSTM 기반 모의유량을 비교한 결과, 전체적으로 모의값이 실측값을 잘 반영하는 것으로 나타났다. 또한, LSTM 모형의 장기간 예측 성능을 평가하기 위하여 LSTM 모형 기반 유량을 보정(2000~2015) 및 검증(2016~2019) 기간의 SWAT 기반 유량에 비교하였다. 비록 모의결과에일부 오차가 발생하였으나, LSTM 모형이 장기간의 하천유량을 잘 산정하는 것으로 나타났다. 검증 결과를 기반으로 2011년~2100년의 CMIP5 미래기후시나리오 기상자료를 이용하여 SWAT 기반 유량을 모의하였으며, 모의한 하천유량을 LSTM 모형의 학습자료로 사용하였다. 다양한 학습 시나리오을 적용하여 LSTM 및 SWAT 모형 기반의 하천유량을 모의하였으며, 최적 학습 기간을 제시하기 위하여 학습 시나리오별 LSTM/SWAT 기반 하천유량의 상관성 및 불확실성을 비교하였다. 비교 결과 학습 기간이 최소 30년 이상일때, 실측유량과 비교하여 LSTM 모형 기반 하천유량의 불확실성이 낮은 것으로 나타났다. 따라서 CMIP5 미래기후시나리오와 딥러닝 기반 LSTM 모형을 연계하여 미래 장기간의 일별 유량을 모의할 경우, 신뢰성 있는 LSTM 모형 기반 하천유량을 모의하기 위해서는 최소 30년 이상의 학습 기간이 필요할 것으로 판단된다.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.36 no.1
• /
• pp.39-48
• /
• 2023

Because a railway bridge is designed over a long period of time and covers a large site, it involves various environmental factors and uncertainties. For this reason, design changes often occur, even if the design was thoroughly reviewed in the initial design stage. In particular, design changes of large-scale facilities, such as railway bridges, consume significant time and cost, and it is extremely inefficient to repeat all the procedures each time. In this study, a technique that can improve the efficiency of learning after design change was developed by utilizing the learning result before design change through transfer learning among deep-learning algorithms. For analysis, scenarios were created, and a database was built using a previously developed railway bridge deep-learning-based prediction system. The proposed method results in similar accuracy when learning only 1000 data points in the new domain compared with the 8000 data points used for learning in the old domain before the design change. Moreover, it was confirmed that it has a faster convergence speed.

• Korean Journal of Remote Sensing
• /
• v.38 no.6_1
• /
• pp.1505-1514
• /
• 2022

Image matching is a crucial preprocessing step for effective utilization of multi-temporal and multi-sensor very high resolution (VHR) satellite images. Deep learning (DL) method which is attracting widespread interest has proven to be an efficient approach to measure the similarity between image pairs in quick and accurate manner by extracting complex and detailed features from satellite images. However, Image matching of VHR satellite images remains challenging due to limitations of DL models in which the results are depending on the quantity and quality of training dataset, as well as the difficulty of creating training dataset with VHR satellite images. Therefore, this study examines the feasibility of DL-based method in matching pair extraction which is the most time-consuming process during image registration. This paper also aims to analyze factors that affect the accuracy based on the configuration of training dataset, when developing training dataset from existing multi-sensor VHR image database with bias for DL-based image matching. For this purpose, the generated training dataset were composed of correct matching pairs and incorrect matching pairs by assigning true and false labels to image pairs extracted using a grid-based Scale Invariant Feature Transform (SIFT) algorithm for a total of 12 multi-temporal and multi-sensor VHR images. The Siamese convolutional neural network (SCNN), proposed for matching pair extraction on constructed training dataset, proceeds with model learning and measures similarities by passing two images in parallel to the two identical convolutional neural network structures. The results from this study confirm that data acquired from VHR satellite image database can be used as DL training dataset and indicate the potential to improve efficiency of the matching process by appropriate configuration of multi-sensor images. DL-based image matching techniques using multi-sensor VHR satellite images are expected to replace existing manual-based feature extraction methods based on its stable performance, thus further develop into an integrated DL-based image registration framework.

• The Journal of Bigdata
• /
• v.6 no.1
• /
• pp.63-81
• /
• 2021

The performance of natural language processing is rapidly improving due to the recent development and application of machine learning and deep learning technologies, and as a result, the field of application is expanding. In particular, as the demand for analysis on unstructured text data increases, interest in NLP(Natural Language Processing) is also increasing. However, due to the complexity and difficulty of the natural language preprocessing process and machine learning and deep learning theories, there are still high barriers to the use of natural language processing. In this paper, for an overall understanding of NLP, by examining the main fields of NLP that are currently being actively researched and the current state of major technologies centered on machine learning and deep learning, We want to provide a foundation to understand and utilize NLP more easily. Therefore, we investigated the change of NLP in AI(artificial intelligence) through the changes of the taxonomy of AI technology. The main areas of NLP which consists of language model, text classification, text generation, document summarization, question answering and machine translation were explained with state of the art deep learning models. In addition, major deep learning models utilized in NLP were explained, and data sets and evaluation measures for performance evaluation were summarized. We hope researchers who want to utilize NLP for various purposes in their field be able to understand the overall technical status and the main technologies of NLP through this paper.