使用映射模型和跨資料集遷移式學習的輕量化居家衰弱症訓練系統;A Lightweight Home-Based Frailty Training System using Mapping Model and Cross-Dataset Transfer Learning

NCU Institutional Repository > 資訊電機學院 > 資訊工程研究所 > 博碩士論文 > Item 987654321/89855

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/89855

题名:	使用映射模型和跨資料集遷移式學習的輕量化居家衰弱症訓練系統;A Lightweight Home-Based Frailty Training System using Mapping Model and Cross-Dataset Transfer Learning
作者:	胡瑄;Hu, Hsuan
贡献者:	資訊工程學系
关键词:	衰弱症;虛擬現實;姿態估算;動作辨識;遷移式學習;Frailty;Virtual Reality;Pose Estimation;Action Recognition;Transfer Learning
日期:	2022-08-01
上传时间:	2022-10-04 12:02:29 (UTC+8)
出版者:	國立中央大學
摘要:	隨著人口老齡化，衰弱症的問題越來越嚴重。許多研究指出，運動可以有效減緩衰弱症狀。與劇烈運動相比，八段錦相當適合衰弱症患者。它由八個簡單的動作組成。許多物理治療師也使用這種常見的氣功來訓練體弱的病人。為了給他們提供更好的訓練方法，本文提出了一種輕量級的基於家庭的衰弱症訓練系統。在系統中，我們設計了一個虛擬八段錦教練。為了達到輕量化的目的，我們使用網路攝影機作為主要設備。該系統還支持 Kinect 框架。我們使用姿勢估計和運動識別方法來分析使用者的運動。除此之外我們提出了一種新的遷移學習方法。我們設計了一個名為“Skeleton Mapnet”的映射模型。其目的是轉換不同框架的骨架數據。該方法使不同框架的數據集能夠共享分類模型。它還可以混合不同框架的骨架數據，解決網路攝影機數據集的不足。這樣的設計提供了系統輕鬆移植到其他平台的能力。此外，該系統還適用於物聯網人工智能的使用。它可以使衰弱症的患者更容易學習和操作。;The problem of frailty is becoming more and more serious with the aging of the population. Many studies have pointed out that exercise can effectively slow down frailty. Compared with vigorous exercise, Baduanjin is quite suitable for frailty patients. It is a traditional Chinese qigong and consists of eight simple movements. Many physical therapists also use this exercise for training frailty patients. To provide them with a better training method, this paper proposes a lightweight family-based frailty training system. In the system, we designed a virtual Baduanjin coach. To achieve the purpose of being lightweight, we use a webcam as the main device. The system also supports the Kinect framework. We use pose estimation and motion recognition methods to analyze the user′s movements. In addition, a novel transfer learning method is proposed. We designed a mapping model called "Skeleton Mapnet". Its purpose is to convert skeleton data of different frameworks. This method enables datasets of different frameworks to share classification models. It can also mix skeleton data of different frameworks to solve the lack of webcam datasets. Such a design provides the ability of the system to be easily ported to other platforms. In addition, the system is also suitable for the use of the Artificial Intelligence of Things. It can make it easier for frailty patients to learn and operate.
显示于类别:	[資訊工程研究所] 博碩士論文

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	17	检视/开启

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....