• Title/Summary/Keyword: multimedia learning

Search Result 1,219, Processing Time 0.022 seconds

Optimization of Action Recognition based on Slowfast Deep Learning Model using RGB Video Data (RGB 비디오 데이터를 이용한 Slowfast 모델 기반 이상 행동 인식 최적화)

  • Jeong, Jae-Hyeok;Kim, Min-Suk
    • Journal of Korea Multimedia Society
    • /
    • v.25 no.8
    • /
    • pp.1049-1058
    • /
    • 2022
  • HAR(Human Action Recognition) such as anomaly and object detection has become a trend in research field(s) that focus on utilizing Artificial Intelligence (AI) methods to analyze patterns of human action in crime-ridden area(s), media services, and industrial facilities. Especially, in real-time system(s) using video streaming data, HAR has become a more important AI-based research field in application development and many different research fields using HAR have currently been developed and improved. In this paper, we propose and analyze a deep-learning-based HAR that provides more efficient scheme(s) using an intelligent AI models, such system can be applied to media services using RGB video streaming data usage without feature extraction pre-processing. For the method, we adopt Slowfast based on the Deep Neural Network(DNN) model under an open dataset(HMDB-51 or UCF101) for improvement in prediction accuracy.

Low-Light Invariant Video Enhancement Scheme Using Zero Reference Deep Curve Estimation (Zero Deep Curve 추정방식을 이용한 저조도에 강인한 비디오 개선 방법)

  • Choi, Hyeong-Seok;Yang, Yoon Gi
    • Journal of Korea Multimedia Society
    • /
    • v.25 no.8
    • /
    • pp.991-998
    • /
    • 2022
  • Recently, object recognition using image/video signals is rapidly spreading on autonomous driving and mobile phones. However, the actual input image/video signals are easily exposed to a poor illuminance environment. A recent researches for improving illumination enable to estimate and compensate the illumination parameters. In this study, we propose VE-DCE (video enhancement zero-reference deep curve estimation) to improve the illumination of low-light images. The proposed VE-DCE uses unsupervised learning-based zero-reference deep curve, which is one of the latest among learning based estimation techniques. Experimental results show that the proposed method can achieve the quality of low-light video as well as images compared to the previous method. In addition, it can reduce the computational complexity with respect to the existing method.

A Study on the Constructivist Multimedia-Assisted Instruction in Secondary School Geography (중등 지리과에서의 구성주의적 멀티미디어 활용 수업의 모형 개발과 효과 분석)

  • Bae, Sang-Woon;Jo, Wha-Ryong
    • Journal of the Korean association of regional geographers
    • /
    • v.5 no.1
    • /
    • pp.163-185
    • /
    • 1999
  • The purpose of this study is to develop the model of constructivist multimedia-assisted instruction(CMAI) and to analyze the effect of it in the secondary school geography. The main results are as follows : (1) The conceptual model of CMAI can be defined as an instruction aiming at making a person who has self-directed learning ability through constructivism and multimedia. The procedural model of CMAI based on PIDA instructional strategy is divided into four stages : prediction & explanation, inquiry activity, discussion & fixation, application & synthesis stage. (2) CMAI is typed by offline CMAI and online CMAI. that is, O/WCMAI(online CMAI by web-based courseware). Offline CMAI is subdivided into P/TCMAI(offline CMAI by presentation-based courseware) and C/RCMAI(offline CMAI by cd-rom based courseware) according to authoring tool and function. (3) Offline constructivist multimedia course-ware(offline courseware) was developed for 2 periods as the material to analyze the effect of CMAI. Offline courseware is received development level of it. (4) After offline courseware being applied to the class, the effect of it according the types of the CMAI instruction(lecture instruction, whole teaching, individualized learning, cooperative learning) was analyzed. As the result of analyzing the descriptive statistics of the level of learning achievement and instruction response, there isn't big relationship between them. As the result of analyzing the inferential statistics of the level of learning achievement, there wasn't significant difference between the types of CMAI instruction in whole student of the classes and certain students who improved their grades. But as the result of analyzing of the level of instruction response, there was significant difference between lecture instruction and other types of the CMAI instruction(whole teaching, individualized learning, cooperative learning).

  • PDF

Optimization of Deep Learning Model Using Genetic Algorithm in PET-CT Image Alzheimer's Classification (PET-CT 영상 알츠하이머 분류에서 유전 알고리즘 이용한 심층학습 모델 최적화)

  • Lee, Sanghyeop;Kang, Do-Young;Song, Jongkwan;Park, Jangsik
    • Journal of Korea Multimedia Society
    • /
    • v.23 no.9
    • /
    • pp.1129-1138
    • /
    • 2020
  • The performance of convolutional deep learning networks is generally determined according to parameters of target dataset, structure of network, convolution kernel, activation function, and optimization algorithm. In this paper, a genetic algorithm is used to select the appropriate deep learning model and parameters for Alzheimer's classification and to compare the learning results with preliminary experiment. We compare and analyze the Alzheimer's disease classification performance of VGG-16, GoogLeNet, and ResNet to select an effective network for detecting AD and MCI. The simulation results show that the network structure is ResNet, the activation function is ReLU, the optimization algorithm is Adam, and the convolution kernel has a 3-dilated convolution filter for the accuracy of dementia medical images.

Stress Identification and Analysis using Observed Heart Beat Data from Smart HRM Sensor Device

  • Pramanta, SPL Aditya;Kim, Myonghee;Park, Man-Gon
    • Journal of Korea Multimedia Society
    • /
    • v.20 no.8
    • /
    • pp.1395-1405
    • /
    • 2017
  • In this paper, we analyses heart beat data to identify subjects stress state (binary) using heart rate variability (HRV) features extracted from heart beat data of the subjects and implement supervised machine learning techniques to create the mental stress classifier. There are four steps need to be done: data acquisition, data processing (HRV analysis), features selection, and machine learning, before doing performance measurement. There are 56 features generated from the HRV Analysis module with several of them are selected (using own algorithm) after computing the Pearson Correlation Matrix (p-values). The results of the list of selected features compared with all features data are compared by its model error after training using several machine learning techniques: support vector machine, decision tree, and discriminant analysis. SVM model and decision tree model with using selected features shows close results compared to using all recording by only 1% difference. Meanwhile, the discriminant analysis differs about 5%. All the machine learning method used in this works have 90% maximum average accuracy.

A Comparison of Distance Metric Learning Methods for Face Recognition (얼굴인식을 위한 거리척도학습 방법 비교)

  • Suvdaa, Batsuri;Ko, Jae-Pil
    • Journal of Korea Multimedia Society
    • /
    • v.14 no.6
    • /
    • pp.711-718
    • /
    • 2011
  • The k-Nearest Neighbor classifier that does not require a training phase is appropriate for a variable number of classes problem like face recognition, Recently distance metric learning methods that is trained with a given data set have reported the significant improvement of the kNN classifier. However, the performance of a distance metric learning method is variable for each application, In this paper, we focus on the face recognition and compare the performance of the state-of-the-art distance metric learning methods, Our experimental results on the public face databases demonstrate that the Mahalanobis distance metric based on PCA is still competitive with respect to both performance and time complexity in face recognition.

CBIR-based Data Augmentation and Its Application to Deep Learning (CBIR 기반 데이터 확장을 이용한 딥 러닝 기술)

  • Kim, Sesong;Jung, Seung-Won
    • Journal of Broadcast Engineering
    • /
    • v.23 no.3
    • /
    • pp.403-408
    • /
    • 2018
  • Generally, a large data set is required for learning of deep learning. However, since it is not easy to create large data sets, there are a lot of techniques that make small data sets larger through data expansion such as rotation, flipping, and filtering. However, these simple techniques have limitation on extendibility because they are difficult to escape from the features already possessed. In order to solve this problem, we propose a method to acquire new image data by using existing data. This is done by retrieving and acquiring similar images using existing image data as a query of the content-based image retrieval (CBIR). Finally, we compare the performance of the base model with the model using CBIR.

A Cascade-hybrid Recommendation Algorithm based on Collaborative Deep Learning Technique for Accuracy Improvement and Low Latency

  • Lee, Hyun-ho;Lee, Won-jin;Lee, Jae-dong
    • Journal of Korea Multimedia Society
    • /
    • v.23 no.1
    • /
    • pp.31-42
    • /
    • 2020
  • During the 4th Industrial Revolution, service platforms utilizing diverse contents are emerging, and research on recommended systems that can be customized to users to provide quality service is being conducted. hybrid recommendation systems that provide high accuracy recommendations are being researched in various domains, and various filtering techniques, machine learning, and deep learning are being applied to recommended systems. However, in a recommended service environment where data must be analyzed and processed real time, the accuracy of the recommendation is important, but the computational speed is also very important. Due to high level of model complexity, a hybrid recommendation system or a Deep Learning-based recommendation system takes a long time to calculate. In this paper, a Cascade-hybrid recommended algorithm is proposed that can reduce the computational time while maintaining the accuracy of the recommendation. The proposed algorithm was designed to reduce the complexity of the model and minimize the computational speed while processing sequentially, rather than using existing weights or using a hybrid recommendation technique handled in parallel. Therefore, through the algorithms in this paper, contents can be analyzed and recommended effectively and real time through services such as SNS environments or shared economy platforms.

Punching Motion Generation using Reinforcement Learning and Trajectory Search Method (경로 탐색 기법과 강화학습을 사용한 주먹 지르기동작 생성 기법)

  • Park, Hyun-Jun;Choi, WeDong;Jang, Seung-Ho;Hong, Jeong-Mo
    • Journal of Korea Multimedia Society
    • /
    • v.21 no.8
    • /
    • pp.969-981
    • /
    • 2018
  • Recent advances in machine learning approaches such as deep neural network and reinforcement learning offer significant performance improvements in generating detailed and varied motions in physically simulated virtual environments. The optimization methods are highly attractive because it allows for less understanding of underlying physics or mechanisms even for high-dimensional subtle control problems. In this paper, we propose an efficient learning method for stochastic policy represented as deep neural networks so that agent can generate various energetic motions adaptively to the changes of tasks and states without losing interactivity and robustness. This strategy could be realized by our novel trajectory search method motivated by the trust region policy optimization method. Our value-based trajectory smoothing technique finds stably learnable trajectories without consulting neural network responses directly. This policy is set as a trust region of the artificial neural network, so that it can learn the desired motion quickly.

Improvement of Activity Recognition Based on Learning Model of AI and Wearable Motion Sensors (웨어러블 동작센서와 인공지능 학습모델 기반에서 행동인지의 개선)

  • Ahn, Junguk;Kang, Un Gu;Lee, Young Ho;Lee, Byung Mun
    • Journal of Korea Multimedia Society
    • /
    • v.21 no.8
    • /
    • pp.982-990
    • /
    • 2018
  • In recent years, many wearable devices and mobile apps related to life care have been developed, and a service for measuring the movement during walking and showing the amount of exercise has been provided. However, they do not measure walking in detail, so there may be errors in the total calorie consumption. If the user's behavior is measured by a multi-axis sensor and learned by a machine learning algorithm to recognize the kind of behavior, the detailed operation of walking can be autonomously distinguished and the total calorie consumption can be calculated more than the conventional method. In order to verify this, we measured activities and created a model using a machine learning algorithm. As a result of the comparison experiment, it was confirmed that the average accuracy was 12.5% or more higher than that of the conventional method. Also, in the measurement of the momentum, the calorie consumption accuracy is more than 49.53% than that of the conventional method. If the activity recognition is performed using the wearable device and the machine learning algorithm, the accuracy can be improved and the energy consumption calculation accuracy can be improved.