An Enhancement of Deep Positional Attention-Based Human Action Recognition by Using Geometric Positional Features

N Srilakshmi; N Radha

doi:10.17485/IJST/v16i29.379

Article

An Enhancement of Deep Positional Attention-Based Human Action Recognition by Using Geometric Positional Features

VIEWS 455
PDF 115

Indian Journal of Science and Technology

DOI: 10.17485/IJST/v16i29.379

Year: 2023, Volume: 16, Issue: 29, Pages: 2190-2197

Original Article

An Enhancement of Deep Positional Attention-Based Human Action Recognition by Using Geometric Positional Features

N Srilakshmi^1*, N Radha²

¹Ph.D. Scholar, Department of Computer Science, PSGR Krishnammal College for Women, Tamilnadu, Coimbatore, India
²Associate Professor, Department of Computer Science, PSGR Krishnammal College for Women, Tamilnadu, Coimbatore, India

*Corresponding Author
Email: [email protected]

Received Date:20 February 2023, Accepted Date:20 June 2023, Published Date:03 August 2023

This work is licensed under a Creative Commons Attribution 4.0 International License.

Abstract

Objective: To learn different geometric features of body joints from video frames, as well as trajectory point coordinates, for Human Activity Recognition (HAR). Methods: Joints and Trajectory-pooled 3D-Deep Geometric Positional Attention-based Hierarchical Bidirectional Recurrent convolutional Descriptors (JTDGPAHBRD)-based HAR framework is proposed. This framework considers the skeleton graph to extract geometric features such as joints, edges, and surfaces, along with the trajectory point coordinates. A new 3D-deep convolutional network with View Conversion (VC) and Temporal Dropout (TD) layers is designed that uses a Positional Attention-based Hierarchical Bidirectional Recurrent Neural Network (PAHBRNN) to learn more discriminatory high-level features. Then, a Fully Connected Layer (FCL) is applied to get the Video Descriptor (VD) of a particular frame. Moreover, the obtained VD is classified by the Support Vector Machine (SVM) classifier to recognize various kinds of human activities. Findings: The test findings show that the JTDGPAHBRD framework using the Penn Action database achieves a recognition rate of 99.7% compared to the existing HAR frameworks. Novelty: This framework has significantly improved the recognition of human activities. Thus, it represents a promising framework for the HAR.

Keywords: Human activity recognition; JTDPAHBRD; Geometric features; View conversion; Temporal dropout; SVM

References

Deshpnande A, Warhade KK. An Improved Model for Human Activity Recognition by Integrated feature Approach and Optimized SVM. In: 2021 International Conference on Emerging Smart Computing and Informatics (ESCI). (pp. 571-576) IEEE. 2021.
Weiyao X, Muqing W, Min Z, Ting XZ. Fusion of Skeleton and RGB Features for RGB-D Human Action Recognition. IEEE Sensors Journal. 2021;21(17):19157–19164. Available from: https://doi.10.1109/JSEN.2021.3089705
Muhammad K, Mustaqeem, Ullah A, Imran AS, Sajjad M, Kiran MS, et al. Human action recognition using attention based LSTM network with dilated CNN features. Future Generation Computer Systems. 2021;125:820–830. Available from: https://doi.org/10.1016/j.future.2021.06.045
Khan S, Khan MA, Alhaisoni M, Tariq U, Yong HSS, Armghan A, et al. Human Action Recognition: A Paradigm of Best Deep Learning Features Selection and Serial Based Extended Fusion. Sensors. 2021;21(23):7941. Available from: https://doi.org/10.3390/s21237941
Wang H, Yu B, Xia K, Li J, Zuo X. Skeleton edge motion networks for human action recognition. Neurocomputing. 2021;423:1–12. Available from: https://doi.org/10.1016/j.neucom.2020.10.037
Saleem R, Ahmad T, Aslam M, Martinez-Enriquez AM. An Intelligent Human Activity Recognizer for Visually Impaired People Using VGG-SVM Model. In: Advances in Computational Intelligence. (Vol. 13613, pp. 356-368) Springer Nature Switzerland. 2022.
Yadav SK, Tiwari K, Pandey HM, Akbar SA. Skeleton-based human activity recognition using ConvLSTM and guided feature learning. Soft Computing. 2022;26(2):877–890. Available from: https://doi.10.1007/s00500-021-06238-7
Putra PU, Shima K, Shimatani K. A deep neural network model for multi-view human activity recognition. PLOS ONE. 2022;17(1):e0262181. Available from: https://doi.org/10.1371/journal.pone.0262181
Li J, Xie Z, Wang Z, Lin Z, Lu C, Zhao Z, et al. A triboelectric gait sensor system for human activity recognition and user identiﬁcation. Nano Energy. 2023;112. Available from: https://doi.org/10.1016/j.nanoen.2023.108473
Nagarathinam S, Narayanan R. Deep Positional Attention-Based Hierarchical Bidirectional RNN with CNN-Based Video Descriptors for Human Action Recognition. International Journal of Intelligent Engineering & Systems. 2022;15(3):406–415. Available from: https://doi:10.22266/ijies2022.0630.34
Srilakshmi N, Radha N. Body Joints and Trajectory Guided 3D Deep Convolutional Descriptors for Human Activity Identification. International Journal of Innovative Technology and Exploring Engineering. 2019;8(12):1016–1021. Available from: https://doi.10.35940/ijitee.K1985.1081219
Srilakshmi N, Radha N. Deep Positional Attention-based Bidirectional RNN with 3D Convolutional Video Descriptors for Human Action Recognition. IOP Conference Series: Materials Science and Engineering. 2021;1022(1):1–10. Available from: https://doi.10.1088/1757-899X/1022/1/012017

Copyright

© 2023 Srilakshmi & Radha. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Published By Indian Society for Education and Environment (iSee)