Implementation of Fall Detection System Based on 3D Skeleton for Deep Learning Technique
In recent years, the fall detection system has become an important topic in the homecare system. Compared with the traditional fall detection algorithm, the method used by neural network is more robust and has higher accuracy. However neural network consumes a large amount of energy due to a huge nu...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2019-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/8869737/ |
id |
doaj-8b58ec55b8164585b29b089d6a31d400 |
---|---|
record_format |
Article |
spelling |
doaj-8b58ec55b8164585b29b089d6a31d4002021-03-29T23:17:09ZengIEEEIEEE Access2169-35362019-01-01715304915305910.1109/ACCESS.2019.29475188869737Implementation of Fall Detection System Based on 3D Skeleton for Deep Learning TechniqueTsung-Han Tsai0https://orcid.org/0000-0001-7524-0621Chin-Wei Hsu1Department of Electrical Engineering, National Central University, Taoyuan City, TaiwanDepartment of Electrical Engineering, National Central University, Taoyuan City, TaiwanIn recent years, the fall detection system has become an important topic in the homecare system. Compared with the traditional fall detection algorithm, the method used by neural network is more robust and has higher accuracy. However neural network consumes a large amount of energy due to a huge number of computations, and needs more memory to store parameters as compared to traditional algorithms. In this paper, we propose a fall detection system in combination of the traditional algorithm with the neural network. First, we propose a skeleton information extraction algorithm, which transforms depth information into skeleton information and extracts the important joints related to fall activity. Also we have modified the skeleton-based method with seven highlight feature points. Second, we propose a highly robust deep convolution neural network architecture, which uses a pruning method to reduce parameters and calculations in the network. The low number of parameters and calculations makes the system suitable for the implementation on an embedded system. The experiment results show the high accuracy and robustness on the popular benchmark dataset NTU RGB+D. The proposed system has been implemented on NVIDIA Jetson Tx2 platform with real-time processing.https://ieeexplore.ieee.org/document/8869737/3D skeletonaction recognitiondeep learningfall detectionembedded systemimage processing |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Tsung-Han Tsai Chin-Wei Hsu |
spellingShingle |
Tsung-Han Tsai Chin-Wei Hsu Implementation of Fall Detection System Based on 3D Skeleton for Deep Learning Technique IEEE Access 3D skeleton action recognition deep learning fall detection embedded system image processing |
author_facet |
Tsung-Han Tsai Chin-Wei Hsu |
author_sort |
Tsung-Han Tsai |
title |
Implementation of Fall Detection System Based on 3D Skeleton for Deep Learning Technique |
title_short |
Implementation of Fall Detection System Based on 3D Skeleton for Deep Learning Technique |
title_full |
Implementation of Fall Detection System Based on 3D Skeleton for Deep Learning Technique |
title_fullStr |
Implementation of Fall Detection System Based on 3D Skeleton for Deep Learning Technique |
title_full_unstemmed |
Implementation of Fall Detection System Based on 3D Skeleton for Deep Learning Technique |
title_sort |
implementation of fall detection system based on 3d skeleton for deep learning technique |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2019-01-01 |
description |
In recent years, the fall detection system has become an important topic in the homecare system. Compared with the traditional fall detection algorithm, the method used by neural network is more robust and has higher accuracy. However neural network consumes a large amount of energy due to a huge number of computations, and needs more memory to store parameters as compared to traditional algorithms. In this paper, we propose a fall detection system in combination of the traditional algorithm with the neural network. First, we propose a skeleton information extraction algorithm, which transforms depth information into skeleton information and extracts the important joints related to fall activity. Also we have modified the skeleton-based method with seven highlight feature points. Second, we propose a highly robust deep convolution neural network architecture, which uses a pruning method to reduce parameters and calculations in the network. The low number of parameters and calculations makes the system suitable for the implementation on an embedded system. The experiment results show the high accuracy and robustness on the popular benchmark dataset NTU RGB+D. The proposed system has been implemented on NVIDIA Jetson Tx2 platform with real-time processing. |
topic |
3D skeleton action recognition deep learning fall detection embedded system image processing |
url |
https://ieeexplore.ieee.org/document/8869737/ |
work_keys_str_mv |
AT tsunghantsai implementationoffalldetectionsystembasedon3dskeletonfordeeplearningtechnique AT chinweihsu implementationoffalldetectionsystembasedon3dskeletonfordeeplearningtechnique |
_version_ |
1724189797427183616 |