| Author: | Yu, Xinbo |
| Title: | Using Kinect v2 network for human activity recognition with application for elderly care |
| Advisors: | Chan, Keith C. C. (COMP) |
| Degree: | M.Sc. |
| Year: | 2015 |
| Subject: | Kinect (Programmable controller) Human activity recognition. Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Computing |
| Pages: | ix, 51 leaves : illustrations (some color) ; 30 cm |
| Language: | English |
| Abstract: | Population aging is increasingly serious as a worldwide problem which affects the global economic development and healthcare expenditure. Developing solutions that support the elderly people to live independently is extremely urgent for the current population aging situation. Recognizing the elderly people's daily activities like fall down, leave home, walk and sleep etc. would improve the security and reliability for elderly people to live independently. Since Microsoft released the Kinect sensor, the computer vision has been experiencing great progress. Utilization of the Kinect sensor in activity recognition for elderly care attracts donation of thousands of researchers and professionals. This thesis firstly investigates the state of the art of utilization of the Kinect sensor in activity recognition systems and validates its potential utilization and development. The state of the arts of activity recognition systems apply various algorithm to process both raw RDB-D data and skeleton data acquired from the Kinect sensor. However, it remains a gap between most of current systems and real world usage. Using a Kinect network to track the whole living area of elderly people is very seldom attempted. Implementation of Kinect sensor network in an elderly person's house would be a potential solution for elderly and bridge the gap, which will truly relieve the effect of population aging if widely applied. In order to validate the effectiveness of Kinect network for activity recognition for elderly care, a single user activity recognition system with three Kinect v2 sensors is developed. In the system, three PCs will host three Kinect v2 sensors separately with one of the PC being a gateway to the WAN and the host PCs will communicate through TCP/IP sockets in a Wi-Fi network. Activities like leave a specific room, walk, sleep, sit down and fall down are detected in real-time. When an emergency action like fall down occur, an email will be sent to other family members' mobile phone. The indoor daily routine of elderly people will be detected based on appearance in predefined locations, which would be further utilized for higher levels of activity recognition and health analysis. To explore the potential solution for expanding the system to one with multiple users, a resident identification solution is proposed for enabling the system to separate the users intelligently. I implement the solution by a SVM classifier. The result turns out that it would be a feasible solution for household usage when users are feature noticeable. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| b27800209.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 2 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/7842