Convolutional Neural Network (CNN) based Gait Recognition System using Microsoft Kinect Skeleton Features

  • Abstract
  • Keywords
  • References
  • PDF
  • Abstract

    Biometric identification systems have recently made exponential advancements in term of complexity and accuracy in recognition for security purposes and a variety of other application. In this paper, a Convolutional Neural Network (CNN) based gait recognition system using Microsoft Kinect skeletal joint data points is proposed for human identification. A total of 23 subjects were used for the experiments. The subjects were positioned 45 degrees (oblique view) from Kinect. A CNN based on the modified AlexNet structure was used to fit the different input data size. The results indicate that the training and testing accuracies were 100% and 69.6% respectively.



  • Keywords

    Convolution Neural Network, biometrics, human gait recognition, Kinect, skeletal joints.

  • References

      [1] Hinton, G. E., Osindero, S., & Teh, Y. W. (2006). A fast learning algorithm for deep belief nets. Neural Computation, 18(7), 1527-1554.

      [2] Arun, P. S. Normal gait cycle.

      [3] Stephanie, C. The Kinect sensor.

      [4] Lawrence, S., Giles, C. L., Tsoi, A. C., & Back, A. D. (1997). Face recognition: A convolutional neural-network approach. IEEE Transactions on Neural Networks, 8(1), 98-113.

      [5] Wang, Y., See, J., Phan, R. C. W., & Oh, Y. H. (2014). Lbp with six intersection points: Reducing redundant information in lbp-top for micro-expression recognition. Proceedings of the Asian Conference on Computer Vision, pp. 525-537.

      [6] Sinha, A., Chakravarty, K., & Bhowmick, B. (2013). Person identification using skeleton information from kinect. Proceedings of the International Conference on Advances in Computer-Human Interactions, pp. 101-108.

      [7] Sahak, R., Yassin, I., Tahir, N. M., Zaman, F. H. K., & Zabidi, A. (2017). Recognition of human gait in oblique and frontal views using Kinect. Journal of Fundamental and Applied Sciences, 9(4S), 1-18.

      [8] MathWorks Inc. 2-D convolutional layer - MATLAB.

      [9] Ioffe, S., & Szegedy, C. (2015). Batch normalization: Accelerating deep network training by reducing internal covariate shift.

      [10] MathWorks Inc. Rectified Linear Unit (ReLU) layer - MATLAB.

      [11] MathWorks Inc. Max pooling layer - MATLAB. pooling&s_tid=doc_srchtitle.

      [12] MathWorks Inc. Fully connected layer - MATLAB.




Article ID: 20806
DOI: 10.14419/ijet.v7i4.11.20806

Copyright © 2012-2015 Science Publishing Corporation Inc. All rights reserved.