• 한국정보통신학회논문지
• /
• 제24권11호
• /
• pp.1403-1410
• /
• 2020

최근, 비대면 경험 및 서비스에 관한 관심이 증가하면서 스마트폰이나 태블릿과 같은 모바일 기기를 이용하여 손쉽게 이용할 수 있는 웹 동영상 콘텐츠에 대한 수요가 급격히 증가하고 있다. 이와 같은 요구사항에 대응하기 위하여, 본 논문에서는 애니메이션이나 영화에 등장하는 명소를 방문하는 무대 탐방 경험을 제공할 수 있는 영상 콘텐츠를 보다 효율적으로 제작하기 위한 기법을 제안한다. 이를 위하여, Google Maps와 Google Street View API를 이용하여 무대탐방 지역에 해당하는 이미지를 수집하여 이미지 데이터셋을 구축하였다. 그 후, 딥러닝 기반의 style transfer 기술을 접목시켜 애니메이션의 독특한 화풍을 실사 이미지에 적용한 후 동영상화하기 위한 방법을 제시하였다. 마지막으로, 다양한 실험을 통해 제안하는 기법을 이용하여 보다 재미있고 흥미로운 형태의 무대탐방 영상 콘텐츠를 생성할 수 있음을 보였다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제8권2호
• /
• pp.483-503
• /
• 2014

In this paper, human action recognition using pyramid histograms of oriented gradients and collaborative multi-task learning is proposed. First, we accumulate global activities and construct motion history image (MHI) for both RGB and depth channels respectively to encode the dynamics of one action in different modalities, and then different action descriptors are extracted from depth and RGB MHI to represent global textual and structural characteristics of these actions. Specially, average value in hierarchical block, GIST and pyramid histograms of oriented gradients descriptors are employed to represent human motion. To demonstrate the superiority of the proposed method, we evaluate them by KNN, SVM with linear and RBF kernels, SRC and CRC models on DHA dataset, the well-known dataset for human action recognition. Large scale experimental results show our descriptors are robust, stable and efficient, and outperform the state-of-the-art methods. In addition, we investigate the performance of our descriptors further by combining these descriptors on DHA dataset, and observe that the performances of combined descriptors are much better than just using only sole descriptor. With multimodal features, we also propose a collaborative multi-task learning method for model learning and inference based on transfer learning theory. The main contributions lie in four aspects: 1) the proposed encoding the scheme can filter the stationary part of human body and reduce noise interference; 2) different kind of features and models are assessed, and the neighbor gradients information and pyramid layers are very helpful for representing these actions; 3) The proposed model can fuse the features from different modalities regardless of the sensor types, the ranges of the value, and the dimensions of different features; 4) The latent common knowledge among different modalities can be discovered by transfer learning to boost the performance.

• 반도체디스플레이기술학회지
• /
• 제21권4호
• /
• pp.65-70
• /
• 2022

In manufacturing and semiconductor industries, transfer robots increase productivity through accurate and continuous work. Due to the nature of the semiconductor process, there are environments where humans cannot intervene to maintain internal temperature and humidity in a clean room. So, transport robots take responsibility over humans. In such an environment where the manpower of the process is cutting down, the lack of maintenance and management technology of the machine may adversely affect the production, and that's why it is necessary to develop a technology for the machine failure diagnosis system. Therefore, this paper tries to identify various causes of failure of transport robots that are widely used in semiconductor automation, and the Prognostics and Health Management (PHM) method is considered for determining and predicting the process of failures. The robot mainly fails in the driving unit due to long-term repetitive motion, and the core components of the driving unit are motors and gear reducer. A simulation drive unit was manufactured and tested around this component and then applied to 6-axis vertical multi-joint robots used in actual industrial sites. Vibration data was collected for each cause of failure of the robot, and then the collected data was processed through signal processing and frequency analysis. The processed data can determine the fault of the robot by utilizing machine learning algorithms such as SVM (Support Vector Machine) and KNN (K-Nearest Neighbor). As a result, the PHM environment was built based on machine learning algorithms using SVM and KNN, confirming that failure prediction was partially possible.

• 인터넷정보학회논문지
• /
• 제20권5호
• /
• pp.1-8
• /
• 2019

본 논문은 지구온난화로 인하여 수온이 상승되며 증가한 해파리의 피해를 감소하고자 연구를 진행하였다. 해수욕장에서 해파리의 등장은 해파리의 쏘임 사고로 인한 인명피해와 폐장으로 인한 경제적 손실이 발생할 수 있다. 본 논문은 선행 연구들로부터 해파리의 출현 패턴을 머신러닝을 통하여 예측 가능함 확인하였다. SVM을 이용한 해파리 출현 예측 모델 연구를 확대하여 진행하였다. 심층신경망을 이용하여 해파리 출현 유무 예측인 이진 분류에서 지수화 된 방법인 다중 분류로 확장하고자 하였다. 수집된 데이터의 크기가 작다는 한계점으로 인하여 84.57%라는 예측 정확도의 한계를 부트스트래핑을 이용하여 데이터 확장을 통해 해결하고자 하였다. 확장된 데이터는 원본 데이터보다 약 7% 이상의 높은 성능을 보여주었으며, Transfer learning과 비교하여 약 6% 이상의 좋은 성능을 보여주었다. 최종적으로 테스트 데이터를 통하여 해파리 출현 예측 성능을 확인한 결과, 해파리의 출현 유무를 예측할 시 높은 정확도로 예측이 가능함을 확인하였으나, 지수화를 통한 예측에서는 의미 있는 결과를 얻지 못하였다.

• 한국멀티미디어학회논문지
• /
• 제25권8호
• /
• pp.1224-1232
• /
• 2022

Whole body bone scan is the most frequently performed nuclear medicine imaging to evaluate bone metastasis in cancer patients. We evaluated the performance of a VGG16-based transfer learning classifier for bone scan images in which metastatic bone lesion was present. A total of 1,000 bone scans in 1,000 cancer patients (500 patients with bone metastasis, 500 patients without bone metastasis) were evaluated. Bone scans were labeled with abnormal/normal for bone metastasis using medical reports and image review. Subsequently, gradient-weighted class activation maps (Grad-CAMs) were generated for explainable AI. The proposed model showed AUROC 0.96 and F1-Score 0.90, indicating that it outperforms to VGG16, ResNet50, Xception, DenseNet121 and InceptionV3. Grad-CAM visualized that the proposed model focuses on hot uptakes, which are indicating active bone lesions, for classification of whole body bone scan images with bone metastases.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2003년도 Proceedings of ACRS 2003 ISRS
• /
• pp.210-212
• /
• 2003

Active sensors are rapidly conquering a share on the remote sensing market and offer among others new possibilities toward automatically acquiring 3D building data. Better dissemination of information about new technological developments can possibly be achieved by short distance-learning courses. The paper describes the didactic and technical aspects of a course we have designed and conducted on airborne laser scanning and interferometric SAR. The building extraction application is a good example to illustrated the added value of short electronic-learning courses above simply publishing (digital) papers.

• 전기전자학회논문지
• /
• 제27권3호
• /
• pp.313-318
• /
• 2023

본 논문은 Vision Transformer를 기반으로 하는 Video Classification의 성능을 개선하는 방법으로 fine-tuning를 적용한 신경망을 제안한다. 최근 딥러닝 기반 실시간 비디오 영상 분석의 필요성이 대두되고 있다. Image Classification에 사용되는 기존 CNN 모델의 특징상 연속된 Frame에 대한 연관성을 분석하기 어렵다는 단점이 있다. 이와 같은 문제를 Attention 메커니즘이 적용된 Vistion Transformer와 Non-local 신경망 모델을 비교 분석하여 최적의 모델을 찾아 해결하고자 한다. 또한, 전이 학습 방법으로 fine-tuning의 다양한 방법을 적용하여 최적의 fine-tuning 신경망 모델을 제안한다. 실험은 UCF101 데이터셋으로 모델을 학습시킨 후, UTA-RLDD 데이터셋에 전이 학습 방법을 적용하여 모델의 성능을 검증하였다.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2022년도 추계학술발표대회
• /
• pp.478-480
• /
• 2022

This paper studies human activity image classification using deep transfer learning techniques focused on the inception convolutional neural networks (InceptionV3) model. For this, we used UFC-101 public datasets containing a group of students' behaviors in mathematics classrooms at a school in Thailand. The video dataset contains Play Sitar, Tai Chi, Walking with Dog, and Student Study (our dataset) classes. The experiment was conducted in three phases. First, it extracts an image frame from the video, and a tag is labeled on the frame. Second, it loads the dataset into the inception V3 with transfer learning for image classification of four classes. Lastly, we evaluate the model's accuracy using precision, recall, F1-Score, and confusion matrix. The outcomes of the classifications for the public and our dataset are 1) Play Sitar (precision = 1.0, recall = 1.0, F1 = 1.0), 2), Tai Chi (precision = 1.0, recall = 1.0, F1 = 1.0), 3) Walking with Dog (precision = 1.0, recall = 1.0, F1 = 1.0), and 4) Student Study (precision = 1.0, recall = 1.0, F1 = 1.0), respectively. The results show that the overall accuracy of the classification rate is 100% which states the model is more powerful for learning UCF-101 and our dataset with higher accuracy.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1989년도 한국자동제어학술회의논문집; Seoul, Korea; 27-28 Oct. 1989
• /
• pp.1069-1074
• /
• 1989

A discrete-time learning control for robotic manipulators is studied using its pulse transfer function. Firstly, discrete-time learning stability condition which is applicable to single-input two-outputs systems is derived. Secondly, stability of learning algorithm with position signal is studied. In this case, when sampling period is small, the algorithm is not stable because of an unstable zero of the system. Thirdly, stability of algorithm with position and velocity signals is studied. In this case, we can stabilize the learning control system which is unstable in learning with only position signal. Finally, simulation results on the trajectory control of robotic manipulators using the discrete-time learning control are shown. This simulation results agree well with the analytical ones.

• 아동학회지
• /
• 제33권2호
• /
• pp.203-222
• /
• 2012

The purpose of this study was to examine the factors influencing their mathematical learning abilities and mathematical learning potential in an attempt to assist their learning at the preschool level. The findings of the study were as follows : First. the female children performed at a much higher level than their male counterparts in terms of mathematical learning ability and mathematical learning potential training. The young children generally improved in their mathematical learning abilities and mathematical learning potential with age. Second, it was found that the participants had higher levels of both mathematical learning ability and mathematical learning potential when their mathematical attitudes and learning motivation were better. Third, there were significant differences in terms training-test and transfer-test scores between the 4 groups based on their relative levels of mathematical abilities and attitudes.