Human Behavioral Action Analysis Using Deep Learning

Authors

  • L. Suresh Professor, Department of Computer Science and Engineering, Cambridge Institute of Technology, Bangalore, India
  • Shreya Nath Student, Department of Computer Science and Engineering, Cambridge Institute of Technology, Bangalore, India
  • Shabana Shaikh Student, Department of Computer Science and Engineering, Cambridge Institute of Technology, Bangalore, India
  • Pratima Kumari Student, Department of Computer Science and Engineering, Cambridge Institute of Technology, Bangalore, India
  • Rashmi Bharti Student, Department of Computer Science and Engineering, Cambridge Institute of Technology, Bangalore, India

Keywords:

Accelerometer data, Convolution Neural Network, Human activity recognition

Abstract

The sensor network-based human activity recognition is an important Fields of study. The work uses methods to manually extract and construct functions of various motions by means of statistical machine learn. Developing Deep Learning technology requires not to extract features manually and improve efficiency in complex issues related to human activities. By migrating the deep neural network image recognition experience, we propose a profound learning model based on Inception Neural Network and recurrent neural networks in combination. The model enters end-to - end data on the waveform of multi-channel sensors. Multi-dimensional features of different kernel-based convergence layers are extracted by Inception related modules. In combination with GRU, time series modelling is carried out and is used fully. Experimental testing on three public HAR datasets that are widely used. In contrast to state of the art, our approach proposed demonstrates consistently superior performance and strong general results.

Downloads

Download data is not yet available.

Downloads

Published

01-08-2020

How to Cite

[1]
L. Suresh, S. . Nath, S. . Shaikh, P. . Kumari, and R. . Bharti, “Human Behavioral Action Analysis Using Deep Learning”, IJRESM, vol. 3, no. 7, pp. 409–412, Aug. 2020.

Issue

Section

Articles