基于空间特征的无线体域网人体姿态识别算法

doi:10.11959/j.issn.2096-3750.2019.00121

物联网学报 ›› 2019, Vol. 3 ›› Issue (3): 70-75.doi: 10.11959/j.issn.2096-3750.2019.00121

基于空间特征的无线体域网人体姿态识别算法

金驰,李志军,孙大洋,胡封晔

吉林大学，吉林长春 130012

修回日期:2019-08-06 出版日期:2019-09-30 发布日期:2019-10-14
作者简介:金驰（1997- ），男，辽宁沈阳人，吉林大学硕士生，主要研究方向为无线体域网和人机交互技术。|李志军（1971- ），男，吉林长春人，吉林大学高级工程师，主要研究方向为无线资源管理技术研究。|孙大洋（1979- ），男，吉林长春人，吉林大学副教授，主要研究方向为无线传感器网络、物联网通信与节能技术、智能终端软件开发。|胡封晔（1974- ），男，河南原阳人，吉林大学教授、博士生导师，主要研究方向为无线体域网、认知无线电、无线能量和信息同传技术、空时通信技术。

Human activity recognition algorithm based on the spatial feature for WBAN

Chi JIN,Zhijun LI,Dayang SUN,Fengye HU

Jilin University,Changchun 130012,China

Revised:2019-08-06 Online:2019-09-30 Published:2019-10-14

摘要/Abstract

摘要：

针对传统基于图像视频的姿态识别算法中所存在的计算成本高、摄像盲区多、隐私易泄露等问题，提出了一种基于手机加速度与陀螺仪数据的卷积—卷积长短时记忆—注意力（CCLA，convolution-convolutional long short-term memory-attention）人体姿态识别算法。使用卷积神经网络对姿态数据进行空间特征提取，采用卷积长短时记忆网络挖掘数据中隐含的时序信息，模拟人脑选择注意力机制构建Attention（注意力）编码器进行更高层次的时空特征提取，以实现对姿态的精准分类。在加州大学欧文分校提出的基于智能手机的人体活动与转换姿态识别数据集上对CCLA算法进行了测试，实现了对12元姿态的分类识别，识别准确率达93.27%。

关键词: 神经网络, 姿态识别, 注意力机制, 无线体域网

Abstract:

Traditional image-based activity recognition algorithms have some problems,such as high computational cost,numerous blind spots and easy privacy leakage.To solve the problem above,the CCLA (convolution-convolutional long short-term memory-attention) activity recognition algorithm based on the acceleration and gyroscope data was proposed.The convolutional neural network was used to extract spatial features of activity data and got the hidden time series information from the convolutional long short-term memory network.Simulating human brain selecting attention mechanism,attention-encoder was constructed to extract the spatial and temporal features at a higher level.The CCLA algorithm was tested on UCI-HAPT (university of California Irvine-smartphone-based recognition of human activities and postural transitions) public data set,and realized the classification of 12 types of activity with the accuracy of 93.27%.

Key words: neural network, activity recognition, attention mechanism, wireless body area network (WBAN)

中图分类号:

TP183

金驰,李志军,孙大洋,胡封晔. 基于空间特征的无线体域网人体姿态识别算法[J]. 物联网学报, 2019, 3(3): 70-75.

Chi JIN,Zhijun LI,Dayang SUN,Fengye HU. Human activity recognition algorithm based on the spatial feature for WBAN[J]. Chinese Journal on Internet of Things, 2019, 3(3): 70-75.

图/表 7

图1

图2

图3

表1

图4

图5

表2

参考文献 19

[1]	VAROL G , LAPTEV I , SCHMID C . Long-term temporal convolutions for action recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017,40(6): 1510-1517.
[2]	HU J F , ZHENG W S , LAI J ,et al. Jointly learning heterogeneous features for RGB-D activity recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017,39(11): 2186-2200.
[3]	DONAHUE J , HENDRICKS L A , ROHRBACH M ,et al. Long-term recurrent convolutional networks for visual recognition and description[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014,39(4): 677-691
[4]	ZHANG L , WU X , LUO D . Human activity recognition with HMM-DNN model[C]// 2015 IEEE 14th International Conference on Cognitive Informatics ＆ Cognitive Computing (ICCI＆CC). IEEE, 2015: 192-197.
[5]	ANGUITA D , GHIO A , ONETO L ,et al. Human activity recognition on smartphones using a multiclass hardware-friendly support vector machine[C]// International Workshop on Ambient Assisted Living. Springer, 2012: 216-223.
[6]	李云鹤 . 智能穿戴设备基于动态模板匹配算法的 3D 手势识别[J]. 物联网学报, 2019,3(1): 97-105.
	LI Y H . 3D gesture recognition based on dynamic template matching algorithm for intelligent wearable devices[J]. Chinese Journal on Internet of Things, 2019,3(1): 97-105.
[7]	ALSHEIKH M A , SELIM A , NIYATO D ,et al. Deep activity recognition models with triaxial accelerometers[C]// Proceedings of the 23rd ACM International Conference on Multimedia. ACM, 2015: 689-692.
[8]	ZENG M , YU T , WANG X ,et al. Semi-supervised convolutional neural networks for human activity recognition[C]// 2017 IEEE International Conference on Big Data (Big Data). IEEE, 2017: 522-529.
[9]	LEE S M , YOON S M , CHO H . Human activity recognition from accelerometer data using convolutional neural network[C]// 2017 IEEE International Conference on Big Data and Smart Computing (BigComp). IEEE, 2017: 131-134.
[10]	JIANG W , YIN Z . Human activity recognition using wearable sensors by deep convolutional neural networks[C]// Proceedings of the 23rd ACM International Conference on Multimedia. ACM, 2015: 1307-1310.
[11]	YANG J , NGUYEN M N , SAN P P ,et al. Deep convolutional neural networks on multichannel time series for human activity recognition[C]// Proceedings of the 24th International Joint Conference on Artificial Intelligence. 2015: 3995-4001.
[12]	HAMMERLA N Y , HALLORAN S , PLOETZ T . Deep,convolutional,and recurrent models for human activity recognition using wearables[J]. Journal of Scientific Computing, 2016,61(2): 454-476.
[13]	SHI X , CHEN Z , WANG H ,et al. Convolutional LSTM network:a machine learning approach for precipitation nowcasting[C]// Advances in Neural Information Processing Systems. 2015: 802-810.
[14]	TAO D , WEN Y , HONG R . Multicolumn bidirectional long short-term memory for mobile devices-based human activity recognition[J]. IEEE Internet of Things Journal, 2016,3(6): 1124-1134.
[15]	ORDó?EZ F , ROGGEN D . Deep convolutional and LSTM recurrent neural networks for multimodal wearable activity recognition[J]. Sensors, 2016,16(1): 115-140.
[16]	RUSH A M , CHOPRA S , WESTON J . A neural attention model for abstractive sentence summarization[C]// Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing. 2015: 379-389.
[17]	LAI J L , YI Y . Key frame extraction based on visual attention model[J]. Journal of Visual Communication and Image Representation, 2012,23(1): 114-125.
[18]	RAFFEL C , ELLIS D P W . Feed-forward networks with attention can solve some long-term memory problems[J]. arXiv, 2016: 1-6.
[19]	REYES-ORTIZ J L , ONETO L , SAMà A ,et al. Transition-aware human activity recognition using smartphones[J]. Neurocomputing, 2016,171: 754-767.

层数	层类型	输出尺寸	参数量/个
1	Conv	(None,8,100,6)	136
2	MaxPooling	(None,8,100,6)	0
3	Conv	(None,8,100,6)	1 032
4	MaxPooling	(None,8,100,6)	0
5	Conv	(None,10,100,6)	1 290
6	Reshape	(None,10,10,10,6)	0
7	Permute	(None,10,10,10,6)	0
8	ConvLSTM	(None,10,8,10,6)	5 216
9	ConvLSTM	(None,10,10,10,6)	6 520
10	Permute	(None,10,10,10,6)	0
11	Reshape	(None,100,60)	0
12	Dense	(None,100,196)	11 956
13	Permute	(None,196,100)	0
14	Dropout	(None,196,100)	0
15	Attention	(None,100)	296
16	Dense	(None,12)	1 212
总参数			27 658

分类器	姿态						准确率
分类器	坐姿	站姿	躺姿	行走	下楼	上楼	准确率
MC-SVM	95.6	97.1	100	87.2	72.6	87.2	89.3%
MCHF-SVM	94.7	97.8	100	87.2	74.0	81.7	89.0%
CCLA	97.0	95.4	97.3	88.9	93.5	98.1	90.2%

基于空间特征的无线体域网人体姿态识别算法

Human activity recognition algorithm based on the spatial feature for WBAN

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 19

相关文章 12

Metrics

推荐阅读 0

[1]	张志飞, 刘峰, 葛祎阳, 李烁, 张煜, 熊轲. 一种基于深度可分离卷积和注意力机制的入侵检测方法[J]. 物联网学报, 2023, 7(1): 49-59.
[2]	苏麟, 党小超, 郝占军, 汝春瑞, 尚旭. 基于WPT-MEC的动态自适应卸载方法[J]. 物联网学报, 2022, 6(4): 128-138.
[3]	郅佳琳, 滕颖蕾, 张新阳, 牛涛, 宋梅. 基于DNN卷积核分割的边缘协作推理性能分析[J]. 物联网学报, 2022, 6(4): 72-81.
[4]	李国权, 徐永海, 林金朝, 黄正文. 基于Res-DNN的端到端MIMO系统信号检测算法[J]. 物联网学报, 2022, 6(1): 65-72.
[5]	杨钦, 杨丽花, 任露露, 黄山虎, 刘加欢. 10 GHz室内无线体域网无线信道测量与建模[J]. 物联网学报, 2022, 6(1): 82-90.
[6]	周毅, 胡姝婷, 李伟, 承楠, 路宁, 沈学民. 图神经网络驱动的交通预测技术：探索与挑战[J]. 物联网学报, 2021, 5(4): 1-16.
[7]	王家君, 曹薇, 张贵龙, 张淮智, 凌子兴, 赵小强. 基于多预测模型与非线性组合的神经网络风速预测[J]. 物联网学报, 2021, 5(4): 81-89.
[8]	赵甘霖, 余畅, 张建富, 杨建新, 冯平法, 沈群. 基于AR虚实图像注意力机制的电缆装配质量检测方法[J]. 物联网学报, 2021, 5(3): 27-38.
[9]	魏泽发,崔华. 基于SqueezeNet卷积神经网络的车辆检测[J]. 物联网学报, 2020, 4(3): 120-125.
[10]	桑永胜,李仁昊,李耀仟,王蔷薇,毛耀. 神经形态视觉传感器及其应用研究[J]. 物联网学报, 2019, 3(4): 63-71.
[11]	于天琪,朱咏絮,王现斌. 基于边缘计算的物联网监测系统中利用自编码神经网络实现的异常检测[J]. 物联网学报, 2018, 2(4): 14-21.
[12]	张雪,石志国,刘璇. 面向不平衡数据的多层神经网络模型[J]. 物联网学报, 2018, 2(2): 65-72.