Human Body Posture Recognition Using Wearable Devices

Recently, the activities of elder people are monitored to support them live independently and safely, where the embedded hardware systems such as wearable devices are widely used. It is a research challenge to deploy deep learning algorithms on embedded devices to recognize the human activities, with the hardware constraints of limited computing resources and low power consumption. In this paper, human body posture recognition methods are proposed for the wearable embedded systems, where back propagation neural network (BPNN) and binary neural network (BNN) are employed to classify the human body postures. The BNN quantizes the synaptic weights and activation values to +1 or −1 based on the BPNN, and is able to achieve a good trade-off between the performance and cost for the embedded systems. In the experiments, the proposed methods are deployed on embedded device of Raspberry Pi 3 for real application of body postures recognition. Results show that compared with BPNN, the BNN can achieve a better trade-off between classification accuracy and cost including required computing resource, power consumption and processing time, e.g. it uses 85.29% less memory, 8.86% less power consumption, and has 5.19% faster classification speed. Therefore, the BNN is more suitable for deployment to resource constrained embedded hardware devices, which is of great significance for the application of human body posture recognition using wearable devices.

[1]  Katarzyna Radecka,et al.  A Comprehensive Analysis on Wearable Acceleration Sensors in Human Activity Recognition , 2017, Sensors.

[2]  H. T. Kung,et al.  Embedded Binarized Neural Networks , 2017, EWSN.

[3]  David Hausheer,et al.  PowerPi: Measuring and modeling the power consumption of the Raspberry Pi , 2014, 39th Annual IEEE Conference on Local Computer Networks.

[4]  Hugo Fuks,et al.  Wearable Computing: Accelerometers' Data Classification of Body Postures and Movements , 2012, SBIA.

[5]  Ig-Jae Kim,et al.  Activity Recognition Using Wearable Sensors for Elder Care , 2008, 2008 Second International Conference on Future Generation Communication and Networking.

[6]  Bogdan Kwolek,et al.  Improving fall detection by the use of depth sensor and accelerometer , 2015, Neurocomputing.

[7]  Christian Desrosiers,et al.  Wearable Devices for Classification of Inadequate Posture at Work Using Neural Networks , 2017, Sensors.

[8]  Song Han,et al.  EIE: Efficient Inference Engine on Compressed Deep Neural Network , 2016, 2016 ACM/IEEE 43rd Annual International Symposium on Computer Architecture (ISCA).

[9]  A. Ng.  Feature selection, L1 vs. L2 regularization, and rotational invariance , 2004, Twenty-first international conference on Machine learning - ICML '04.

[10]  Rainer Stiefelhagen,et al.  CNN-based sensor fusion techniques for multimodal human activity recognition , 2017, SEMWEB.

[11]  Miguel A. Labrador,et al.  A Survey on Human Activity Recognition using Wearable Sensors , 2013, IEEE Communications Surveys & Tutorials.

[12]  D Stephenson,et al.  Infilling streamflow data using feed-forward back-propagation (BP) artificial neural networks: Application of standard BP and pseudo Mac Laurin power series BP techniques , 2007 .

[13]  Wayne Luk,et al.  FP-BNN: Binarized neural network on FPGA , 2018, Neurocomputing.

[14]  Ali Farhadi,et al.  XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks , 2016, ECCV.

[15]  L. Mathew,et al.  Increasing trend of wearables and multimodal interface for human activity monitoring: A review. , 2017, Biosensors & bioelectronics.

[16]  Alexander Wong,et al.  Tiny SSD: A Tiny Single-Shot Detection Deep Convolutional Neural Network for Real-Time Embedded Object Detection , 2018, 2018 15th Conference on Computer and Robot Vision (CRV).

[17]  Misha Denil,et al.  Predicting Parameters in Deep Learning , 2014 .

[18]  Fengye Hu,et al.  A human body posture recognition algorithm based on BP neural network for wireless body area networks , 2016, China Communications.

[19]  Andrés Vázquez Rodas,et al.  Minimizing the power consumption in Raspberry Pi to use as a remote WSN gateway , 2016, 2016 8th IEEE Latin-American Conference on Communications (LATINCOM).

[20]  R. Hubbard,et al.  The ageing of the population: implications for multidisciplinary care in hospital. , 2004, Age and ageing.

[21]  Rosalind W. Picard,et al.  A Wearable Sensor for Unobtrusive, Long-Term Assessment of Electrodermal Activity , 2010, IEEE Transactions on Biomedical Engineering.

[22]  Giuseppe De Pietro,et al.  A Real-time m-Health Monitoring System: An Integrated Solution Combining the Use of Several Wearable Sensors and Mobile Devices , 2017, HEALTHINF.

[23]  James Church,et al.  Wearable sensor badge and sensor jacket for context awareness , 1999, Digest of Papers. Third International Symposium on Wearable Computers.

[24]  Warren J. Gross,et al.  An Architecture to Accelerate Convolution in Deep Neural Networks , 2018, IEEE Transactions on Circuits and Systems I: Regular Papers.

[25]  Huawei Chen,et al.  Compressing Deep Convolutional Networks Using K-means Based on Weights Distribution , 2017, ICIIP.

[26]  Vivienne Sze,et al.  Efficient Processing of Deep Neural Networks: A Tutorial and Survey , 2017, Proceedings of the IEEE.

[27]  Gang Hua,et al.  How to Train a Compact Binary Neural Network with High Accuracy? , 2017, AAAI.

[28]  José Salvador Sánchez,et al.  Surrounding neighborhood-based SMOTE for learning from imbalanced data sets , 2012, Progress in Artificial Intelligence.

[29]  Wei Zhang,et al.  A Unified Framework for Street-View Panorama Stitching , 2016, Sensors.